FairMOT(Fairness of Detection and Re-Identification in Multiple Object Tracking)是一种基于深度学习的多目标跟踪算法,采用目标检测和跟踪相结合的方式,在目标检测的基础上实现了多目标跟踪。它可以在视频中对多个目标进行跟踪,支持长时序的目标跟踪。是一种端到端的目标跟踪算法,它基于目标检测器,利用多任务学习的方法,同时进行目标检测和目标跟踪,以提高跟踪精度和效率。
FairMOT的核心思想是将目标检测和跟踪相结合,使用目标检测算法检测视频帧中的目标,然后通过跟踪算法追踪目标在视频序列中的运动轨迹。在跟踪的过程中,FairMOT采用了一个多任务学习框架,同时预测目标的位置、运动状态和身份信息,这可以提高跟踪的准确性和稳定性。FairMOT的优点是可以在复杂的场景中实现多目标跟踪,支持长时序的跟踪,能够处理目标的遮挡和运动模糊等问题。它在多个数据集上取得了很好的性能,并且具有较高的实时性。
FairMOT使用的是基于卷积神经网络的跟踪器,通过在目标检测器中添加跟踪头来实现目标跟踪,以此提高跟踪的准确性和稳定性。同时,FairMOT还采用了一种基于卡尔曼滤波器的动态权重分配机制,可以对不同目标的跟踪结果进行动态调整,以提高跟踪效果。FairMOT在多目标跟踪方面具有较好的性能,尤其在目标重叠、遮挡、运动模糊等复杂情况下仍能保持较高的跟踪精度和鲁棒性。值得注意的是,FairMOT并不是唯一的多目标跟踪算法,还有很多其他的算法可以实现类似的功能。此外,由于多目标跟踪是一个非常具有挑战性的任务,算法的性能取决于很多因素,包括数据集、算法参数、硬件环境等。因此,在实际应用中需要根据具体情况选择合适的算法。
环境准备
1.PC Host Ubuntu 18.04.6,Linux Kernel 5.4,内核版本关系不大,记录下来备查。
2.安装基础工具,比如GCC,CMAKE,VIM,GIT等等,工具尽量完备, 如果做不到,遇到问题临时下载也可。
3.安装python3发行版,我用的是anaconda发行版,具体版本是
Anaconda3-2020.11-Linux-x86_64.sh
下载地址在如下链接,选择对应的版本即可。
https://repo.anaconda.com/archivehttps://repo.anaconda.com/archive4.对应的Python版本为3.8.5版
FairMOT模型下载:
从以下地址将FairMOT模型下载到本地:
git clone https://github.com/ifzhang/FairMOT.git
安装依赖:
FairMOT模型依赖Python 3.8及以上版本以及pytorch >= 1.7.0,前者条件已经满足,后者需要在这步安装
conda create -n FairMOT
conda activate FairMOT
conda install pytorch==1.7.0 torchvision==0.8.0 cudatoolkit=10.2 -c pytorch
cd FairMOT
pip install cython
pip install -r requirements.txt
执行conda activate FairMOT之后,python会进入一个新的FariMOT环境,命令行提示符前会冠FairMOT字符。如果想退出 此环境,执行:
conda deactivate
conda也支持另一种创建环境的语法,conda env create -f environment.yml,其中 environment.yml是一个描述被创环境的文件,需要符合一定的语法格式,举例如下:
退出FairMOT环境,回到base环境,继续执行上述命令,回到系统环境.
执行conda info --envs可以查看安装了哪些环境:
此步操作完之后,执行conda list查看安装的依赖包情况如下:
(FairMOT) caozilong@caozilong-RedmiBook-14:~$ conda list
# packages in environment at /home/caozilong/anaconda3/envs/FairMOT:
#
# Name Version Build Channel
_libgcc_mutex 0.1 main https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
_openmp_mutex 4.5 1_gnu https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
attrs 21.2.0 pypi_0 pypi
blas 1.0 mkl https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
ca-certificates 2021.10.26 h06a4308_2 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
certifi 2021.10.8 py38h06a4308_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
cudatoolkit 10.2.89 hfd86e86_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
cycler 0.11.0 pypi_0 pypi
cython 0.29.24 pypi_0 pypi
cython-bbox 0.1.3 pypi_0 pypi
dcnv2 0.1 dev_0 <develop>
et-xmlfile 1.1.0 pypi_0 pypi
flake8 4.0.1 pypi_0 pypi
flake8-import-order 0.18.1 pypi_0 pypi
freetype 2.11.0 h70c0345_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
fvcore 0.1.5.post20211023 pypi_0 pypi
giflib 5.2.1 h7b6447c_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
iniconfig 1.1.1 pypi_0 pypi
intel-openmp 2021.4.0 h06a4308_3561 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
iopath 0.1.9 pypi_0 pypi
jpeg 9d h7f8727e_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
kiwisolver 1.3.2 pypi_0 pypi
lap 0.4.0 pypi_0 pypi
lcms2 2.12 h3be6417_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
ld_impl_linux-64 2.35.1 h7274673_9 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libffi 3.3 he6710b0_2 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libgcc-ng 9.3.0 h5101ec6_17 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libgomp 9.3.0 h5101ec6_17 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libpng 1.6.37 hbc83047_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libstdcxx-ng 9.3.0 hd4cf53a_17 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libtiff 4.2.0 h85742a9_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libuv 1.40.0 h7b6447c_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libwebp 1.2.0 h89dd481_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
libwebp-base 1.2.0 h27cfd23_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
lz4-c 1.9.3 h295c915_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
matplotlib 3.4.3 pypi_0 pypi
mccabe 0.6.1 pypi_0 pypi
mkl 2021.4.0 h06a4308_640 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
mkl-service 2.4.0 py38h7f8727e_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
mkl_fft 1.3.1 py38hd3c417c_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
mkl_random 1.2.2 py38h51133e4_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
motmetrics 1.2.0 pypi_0 pypi
ncurses 6.3 heee7806_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
ninja 1.10.2 hff7bd54_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
numpy 1.21.2 py38h20f2e39_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
numpy-base 1.21.2 py38h79a1101_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
olefile 0.46 pyhd3eb1b0_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
opencv-python 4.5.4.58 pypi_0 pypi
openpyxl 3.0.9 pypi_0 pypi
openssl 1.1.1l h7f8727e_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
packaging 21.2 pypi_0 pypi
pandas 1.3.4 pypi_0 pypi
pillow 8.4.0 py38h5aabda8_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pip 21.2.4 py38h06a4308_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
pluggy 1.0.0 pypi_0 pypi
portalocker 2.3.2 pypi_0 pypi
progress 1.6 pypi_0 pypi
py 1.11.0 pypi_0 pypi
py-cpuinfo 8.0.0 pypi_0 pypi
pycodestyle 2.8.0 pypi_0 pypi
pyflakes 2.4.0 pypi_0 pypi
pyparsing 2.4.7 pypi_0 pypi
pytest 6.2.5 pypi_0 pypi
pytest-benchmark 3.4.1 pypi_0 pypi
python 3.8.12 h12debd9_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
python-dateutil 2.8.2 pypi_0 pypi
pytorch 1.7.0 py3.8_cuda10.2.89_cudnn7.6.5_0 pytorch
pytz 2021.3 pypi_0 pypi
pyyaml 6.0 pypi_0 pypi
readline 8.1 h27cfd23_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
scipy 1.7.2 pypi_0 pypi
setuptools 58.0.4 py38h06a4308_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
six 1.16.0 pyhd3eb1b0_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
sqlite 3.36.0 hc218d9a_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
tabulate 0.8.9 pypi_0 pypi
tensorboardx 2.4 pypi_0 pypi
tk 8.6.11 h1ccaba5_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
toml 0.10.2 pypi_0 pypi
torchvision 0.8.0 py38_cu102 pytorch
tqdm 4.62.3 pypi_0 pypi
typing_extensions 3.10.0.2 pyh06a4308_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
wheel 0.37.0 pyhd3eb1b0_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
xmltodict 0.12.0 pypi_0 pypi
xz 5.2.5 h7b6447c_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
yacs 0.1.8 pypi_0 pypi
zlib 1.2.11 h7b6447c_3 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
zstd 1.4.9 haebb681_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
(FairMOT) caozilong@caozilong-RedmiBook-14:~$
requirement.txt中的依赖项如下所示:
根据安装包的列表可以看出,此模型是不依赖于tesorflow的,fairmot是基于Pytorch框架进行训练和推理的。
值得注意的是,根据名字也可以看出,pytorch1.7.0依赖于py3.8和cuda10.2,这里python3.8.12进行了二次安装,而CUDA我们将在下一步介绍其安装方式。
根据下图可以看到,虽然anaconda中已经安装了3.8.5版本的python,但是fairmot环境使用的是上一步安装的3.8.12版本的python,由于没有发现有地方显示的指定重新安装Python的版本,怀疑是Pytorch包本身内建了对3.8.12版本python的的依赖。
根据GITHUB官方介绍:
“We use DCNv2_pytorch_1.7 in our backbone network (pytorch_1.7 branch). Previous versionscan be found in DCNv2.”
所以,我们需要安装DCNV2,执行如下命令安装:
git clone -b pytorch_1.7 https://github.com/ifzhang/DCNv2.git
cd DCNv2
./make.sh
最后,需要安装FFMPEG,FFMPEG不是必须的,他的作用是,在最后推理完成后,将推理后的画框照片合成视频文件用的,安装方法可以从源码安装,也可以用sudo apt install ffmpeg安装。源码安装的方式请参考:
ubuntu18.04编译FFMPEG_tugouxp的专栏-CSDN博客
本文使用的是后者,命令行安装的方式。
CUDA安装:
模型运行依赖于支持较新版本CUDA的Nvidia GPU硬件,所以需要安装N卡驱动以及CUDA. 如果在不支持N卡和CUDA的PC上搭建环境,最终执行模型测试的时候,会遇到下面的问题,血的教训。
在UBUNTU上,N卡驱动默认使用的是nouveau, 不满足需求,需要将其屏蔽,方法是列入黑名单禁用并且根据显卡型号安装正确的驱动版本,之后才能安装CUDA.
这一步本文只作简要记录,主要记录过程中遇到的一些问题以及解决方法,具体安装细节可以参考博客:
Ubuntu18.04安装CUDA深度学习环境_tugouxp的专栏-CSDN博客N卡在深度学习领域具备无可替代的地位,这里记录以下在我这台配备GF MX350 N卡笔记本 上安装cuda以及cuNN的具体步骤。MX350属于低端显卡了,一开始我还担心它会不会不支持CUDA,后面确认了一下,发现可以支持。关于确认N卡是否支持CUDA以及支持版本的具体步骤,可以参考我的这篇博客:如何确定PC Nvidia显卡是否支持CUDA以及cudaNN?过程不难,细节很多,下面记录具体步骤:1.安装显卡驱动:...https://blog.csdn.net/tugouxp/article/details/119829792我的显卡型号是GetForce MX250,是支持较新版本的CUDA的。
CUDA下载地址:CUDA Toolkit Archive | NVIDIA Developer
根据pytorch的要求,选择10.2版本的cuda.
之后经过层层选择之后,下方会列出正确的安装命令:
安装命令:
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-ubuntu1804.pin
sudo mv cuda-ubuntu1804.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/10.2/Prod/local_installers/cuda-repo-ubuntu1804-10-2-local-10.2.89-440.33.01_1.0-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1804-10-2-local-10.2.89-440.33.01_1.0-1_amd64.deb
sudo apt-key add /var/cuda-repo-10-2-local-10.2.89-440.33.01/7fa2af80.pub
sudo apt-get update
sudo apt-get -y install cuda
CUDA deb 文件比较大,如果选择离线方式下载,可以用快一些的下载工具替代wget下载的方式,下载过来后,文件是这样的,有1.8个G。
按照如上的命令下载,不会出错的。
关于下方的PATCH文件是否一定要安装,我的测试来看是不需要的,不安装PATCH文件,也可以跑出正确的结果。
安装的最后两步执行命令sudo apt-get update,sudo apt-get -y install cuda完成后,即可添加环境变量到.bashrc
细心的你一定注意到了,前面我们安装时选择的CUDA安装版本是10.2,怎么安装完成后变成11.5了,关于这个问题,我也不是很确定其中的原因,只是经过测试,用11.5的CUDA也可以正确的跑出FairMOT的结果。版本变化的个中原因,后面在研究。
关于CUDA安装路径和安装方式,可以看到在/usr/local下有三个CUDA安装目录,分别是cuda,cua-11,以及cuda-11.5,在仔细分析发现,前两个均是间接指向cuda-11.5的软链接,所以结论就是无论使用那个工具目录作为入口,使用的其实都是cuda11.5,而并非是安装了三个版本的CUDA.至于为什么这样做,怀疑是为了保证随着CUDA版本的不断升级,毅然能保持兼容各类使用CUDA的开源框架。使CUDA和开源框架之间做到解耦。这有些类似于软件设计中分层的思想。
比如darknet构建框架中对CUDA的引用,它直接使用了抽象的链接目录/usr/local/cuda作为引用目录,这样即使CUDA版本升级,依然不需要改darknet。
安装完成后,重启机器,以便环境变量生效,之后执行nvidia-smi命令查看N卡和CUDA信息:
除此之外,安装好的CUDA环境还提供了NVCC编译器工具,可以将编译代码投入到N卡加速运行:
CUDA安装目录中还提供了很多的测试用例,可以用来对对CUDA环境进行测试,以计算矩阵特征值的用例为例:/usr/local/cuda-11.5/samples/6_Advanced/eigenvalues
查看显卡驱动信息:
进入/usr/local/cuda-11.5/samples/1_Utilities/deviceQuery目录,执行编译:
之后,用生成的./deviceQuery工具运行显卡检测:
可以看到,显卡型号是“NVIDIA GeForce MX250”,有384 CUDA 核心,主频1.58G,内存3504MHZ等信息。
测试用是题外话,和模型验证无关。
下载训练和推理数据集文件:
到网址:
下载训练和推理数据集:
下载5.5G大小的文件,包含所有的验证训练数据:
模型比较大,建议用下载工具下载。
下载后的文件如下图所示:
模型准备:
到以下地址下载两个模型文件
https://drive.google.com/file/d/1Ur3_pa9r3KRY-5qM2cdFhFJ5exghRJvh/view?usp=sharing
https://drive.google.com/file/d/1MEvsRPyoAqYSCdKaS5Ofrl7ZfKbBZ1Jb/view?usp=sharing
如果因为墙的原因下载不了,也可以到以下网盘下载:
提取码分别为:wh9h和2y3a
下载后,在前面下载的FairMOT项目顶层目录中创建一个新的model目录,并将模型文件放进取。
目录位置和名称任意,最终会体现在命令行参数上。
解压训练数据:
将下载的MOT17.ZIP解压到~/dataset/目录,执行
caozilong@caozilong-RedmiBook-14:~/dataset$ unzip MOT17.zip
修改模型路径:
修改~/FairMOT/src/lib/cfg/data.json和~/FairMOT/src/lib/opts.py文件,使其路径指向我们的数据集路径:
执行模型:
conda activate FairMOT 环境下,进入到~/FairMOT/src目录,执行命令:
python track.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
一开始遇到错误:
此错误是由路径引起的,分析track.py代码,发现其路径是有两部分拼接而成,一部分是前面步骤修改的data_dir,后一部分是 images/train,可是从前面解压的目录树情况来看,没有images目录。
修改方式是创建images目录,不过貌似执行过一次错误的运行后,images目录已经被创建了:
我们只需要将 train目录拷贝过来即可,
此时的目录结构如下图所示,和上次比,结构不变,只是image/train下多了内容。
此时,再一次执行命令:
python track.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
发现上面的错误没有了,但是出现了新的错误:
报错提示,没有找到支持CUDA的GPU设备,可是进入交互命令打印GPU数量,明明是找的到CUDA GPU设备的:
无解,后面白嫖百度后得到一个方案,我把它截图下来:
修改track.py文件172行,将1改为0
完成后,当前所有的修改如下所示:
再次执行命令:
python track.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
这次成功执行,之所以产生这种现象,是因为在使用Cuda之前都需要检测一下cuda是否可用,没检测就默认没有,而这里如果设置为非0值,可能会导致代码中不去执行检测,使用的时候也就无法正常使用。
以下是执行过程中全部的LOG:
(FairMOT) caozilong@caozilong-RedmiBook-14:~/FairMOT/src$ python track.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
/home/caozilong/FairMOT/src/lib/models/networks/pose_hrnet.py:523: SyntaxWarning: "is" with a literal. Did you mean "=="?
or self.pretrained_layers[0] is '*':
Fix size testing.
training chunk_sizes: [6, 6]
The output will be saved to /home/caozilong/FairMOT/src/lib/../../exp/mot/default
heads {'hm': 1, 'wh': 4, 'id': 64, 'reg': 2}
2021-11-10 18:29:19 [INFO]: start seq: MOT17-02-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:29:22 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:29:25 [INFO]: Processing frame 20 (9.79 fps)
2021-11-10 18:29:27 [INFO]: Processing frame 40 (11.12 fps)
2021-11-10 18:29:29 [INFO]: Processing frame 60 (11.65 fps)
2021-11-10 18:29:31 [INFO]: Processing frame 80 (11.92 fps)
2021-11-10 18:29:33 [INFO]: Processing frame 100 (12.10 fps)
2021-11-10 18:29:35 [INFO]: Processing frame 120 (12.22 fps)
2021-11-10 18:29:37 [INFO]: Processing frame 140 (12.31 fps)
2021-11-10 18:29:39 [INFO]: Processing frame 160 (12.36 fps)
2021-11-10 18:29:41 [INFO]: Processing frame 180 (12.41 fps)
2021-11-10 18:29:44 [INFO]: Processing frame 200 (12.44 fps)
2021-11-10 18:29:46 [INFO]: Processing frame 220 (12.47 fps)
2021-11-10 18:29:48 [INFO]: Processing frame 240 (12.49 fps)
2021-11-10 18:29:50 [INFO]: Processing frame 260 (12.50 fps)
2021-11-10 18:29:52 [INFO]: Processing frame 280 (12.49 fps)
2021-11-10 18:29:54 [INFO]: Processing frame 300 (12.50 fps)
2021-11-10 18:29:56 [INFO]: Processing frame 320 (12.50 fps)
2021-11-10 18:29:59 [INFO]: Processing frame 340 (12.50 fps)
2021-11-10 18:30:01 [INFO]: Processing frame 360 (12.51 fps)
2021-11-10 18:30:03 [INFO]: Processing frame 380 (12.50 fps)
2021-11-10 18:30:05 [INFO]: Processing frame 400 (12.50 fps)
2021-11-10 18:30:07 [INFO]: Processing frame 420 (12.50 fps)
2021-11-10 18:30:09 [INFO]: Processing frame 440 (12.51 fps)
2021-11-10 18:30:11 [INFO]: Processing frame 460 (12.52 fps)
2021-11-10 18:30:14 [INFO]: Processing frame 480 (12.52 fps)
2021-11-10 18:30:16 [INFO]: Processing frame 500 (12.52 fps)
2021-11-10 18:30:18 [INFO]: Processing frame 520 (12.51 fps)
2021-11-10 18:30:20 [INFO]: Processing frame 540 (12.51 fps)
2021-11-10 18:30:22 [INFO]: Processing frame 560 (12.52 fps)
2021-11-10 18:30:24 [INFO]: Processing frame 580 (12.52 fps)
2021-11-10 18:30:26 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-02-SDP.txt
2021-11-10 18:30:26 [INFO]: Evaluate seq: MOT17-02-SDP
2021-11-10 18:30:27 [INFO]: start seq: MOT17-04-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:30:27 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:30:30 [INFO]: Processing frame 20 (12.12 fps)
2021-11-10 18:30:32 [INFO]: Processing frame 40 (12.13 fps)
2021-11-10 18:30:34 [INFO]: Processing frame 60 (12.13 fps)
2021-11-10 18:30:36 [INFO]: Processing frame 80 (12.10 fps)
2021-11-10 18:30:39 [INFO]: Processing frame 100 (12.11 fps)
2021-11-10 18:30:41 [INFO]: Processing frame 120 (12.12 fps)
2021-11-10 18:30:43 [INFO]: Processing frame 140 (12.11 fps)
2021-11-10 18:30:45 [INFO]: Processing frame 160 (12.12 fps)
2021-11-10 18:30:48 [INFO]: Processing frame 180 (12.13 fps)
2021-11-10 18:30:50 [INFO]: Processing frame 200 (12.13 fps)
2021-11-10 18:30:52 [INFO]: Processing frame 220 (12.12 fps)
2021-11-10 18:30:54 [INFO]: Processing frame 240 (12.12 fps)
2021-11-10 18:30:57 [INFO]: Processing frame 260 (12.10 fps)
2021-11-10 18:30:59 [INFO]: Processing frame 280 (12.09 fps)
2021-11-10 18:31:01 [INFO]: Processing frame 300 (12.08 fps)
2021-11-10 18:31:04 [INFO]: Processing frame 320 (12.08 fps)
2021-11-10 18:31:06 [INFO]: Processing frame 340 (12.07 fps)
2021-11-10 18:31:08 [INFO]: Processing frame 360 (12.06 fps)
2021-11-10 18:31:10 [INFO]: Processing frame 380 (12.06 fps)
2021-11-10 18:31:13 [INFO]: Processing frame 400 (12.06 fps)
2021-11-10 18:31:15 [INFO]: Processing frame 420 (12.06 fps)
2021-11-10 18:31:17 [INFO]: Processing frame 440 (12.05 fps)
2021-11-10 18:31:19 [INFO]: Processing frame 460 (12.04 fps)
2021-11-10 18:31:22 [INFO]: Processing frame 480 (12.04 fps)
2021-11-10 18:31:24 [INFO]: Processing frame 500 (12.04 fps)
2021-11-10 18:31:26 [INFO]: Processing frame 520 (12.03 fps)
2021-11-10 18:31:29 [INFO]: Processing frame 540 (12.03 fps)
2021-11-10 18:31:31 [INFO]: Processing frame 560 (12.03 fps)
2021-11-10 18:31:33 [INFO]: Processing frame 580 (12.03 fps)
2021-11-10 18:31:35 [INFO]: Processing frame 600 (12.02 fps)
2021-11-10 18:31:38 [INFO]: Processing frame 620 (12.03 fps)
2021-11-10 18:31:40 [INFO]: Processing frame 640 (12.02 fps)
2021-11-10 18:31:42 [INFO]: Processing frame 660 (12.02 fps)
2021-11-10 18:31:45 [INFO]: Processing frame 680 (12.02 fps)
2021-11-10 18:31:47 [INFO]: Processing frame 700 (12.02 fps)
2021-11-10 18:31:49 [INFO]: Processing frame 720 (12.02 fps)
2021-11-10 18:31:52 [INFO]: Processing frame 740 (12.02 fps)
2021-11-10 18:31:54 [INFO]: Processing frame 760 (12.01 fps)
2021-11-10 18:31:56 [INFO]: Processing frame 780 (12.01 fps)
2021-11-10 18:31:59 [INFO]: Processing frame 800 (12.01 fps)
2021-11-10 18:32:01 [INFO]: Processing frame 820 (12.01 fps)
2021-11-10 18:32:03 [INFO]: Processing frame 840 (12.00 fps)
2021-11-10 18:32:06 [INFO]: Processing frame 860 (12.00 fps)
2021-11-10 18:32:08 [INFO]: Processing frame 880 (11.99 fps)
2021-11-10 18:32:10 [INFO]: Processing frame 900 (11.99 fps)
2021-11-10 18:32:13 [INFO]: Processing frame 920 (11.98 fps)
2021-11-10 18:32:15 [INFO]: Processing frame 940 (11.98 fps)
2021-11-10 18:32:17 [INFO]: Processing frame 960 (11.98 fps)
2021-11-10 18:32:19 [INFO]: Processing frame 980 (11.97 fps)
2021-11-10 18:32:22 [INFO]: Processing frame 1000 (11.97 fps)
2021-11-10 18:32:24 [INFO]: Processing frame 1020 (11.97 fps)
2021-11-10 18:32:27 [INFO]: Processing frame 1040 (11.96 fps)
2021-11-10 18:32:28 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-04-SDP.txt
2021-11-10 18:32:28 [INFO]: Evaluate seq: MOT17-04-SDP
2021-11-10 18:32:30 [INFO]: start seq: MOT17-05-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:32:30 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:32:32 [INFO]: Processing frame 20 (12.99 fps)
2021-11-10 18:32:34 [INFO]: Processing frame 40 (13.01 fps)
2021-11-10 18:32:36 [INFO]: Processing frame 60 (13.01 fps)
2021-11-10 18:32:38 [INFO]: Processing frame 80 (13.03 fps)
2021-11-10 18:32:40 [INFO]: Processing frame 100 (12.94 fps)
2021-11-10 18:32:41 [INFO]: Processing frame 120 (12.94 fps)
2021-11-10 18:32:43 [INFO]: Processing frame 140 (12.93 fps)
2021-11-10 18:32:45 [INFO]: Processing frame 160 (12.93 fps)
2021-11-10 18:32:47 [INFO]: Processing frame 180 (12.94 fps)
2021-11-10 18:32:49 [INFO]: Processing frame 200 (12.93 fps)
2021-11-10 18:32:51 [INFO]: Processing frame 220 (12.93 fps)
2021-11-10 18:32:53 [INFO]: Processing frame 240 (12.93 fps)
2021-11-10 18:32:54 [INFO]: Processing frame 260 (12.92 fps)
2021-11-10 18:32:56 [INFO]: Processing frame 280 (12.91 fps)
2021-11-10 18:32:58 [INFO]: Processing frame 300 (12.91 fps)
2021-11-10 18:33:00 [INFO]: Processing frame 320 (12.89 fps)
2021-11-10 18:33:02 [INFO]: Processing frame 340 (12.89 fps)
2021-11-10 18:33:04 [INFO]: Processing frame 360 (12.87 fps)
2021-11-10 18:33:06 [INFO]: Processing frame 380 (12.87 fps)
2021-11-10 18:33:08 [INFO]: Processing frame 400 (12.88 fps)
2021-11-10 18:33:09 [INFO]: Processing frame 420 (12.88 fps)
2021-11-10 18:33:11 [INFO]: Processing frame 440 (12.88 fps)
2021-11-10 18:33:13 [INFO]: Processing frame 460 (12.88 fps)
2021-11-10 18:33:15 [INFO]: Processing frame 480 (12.89 fps)
2021-11-10 18:33:17 [INFO]: Processing frame 500 (12.89 fps)
2021-11-10 18:33:18 [INFO]: Processing frame 520 (12.89 fps)
2021-11-10 18:33:20 [INFO]: Processing frame 540 (12.89 fps)
2021-11-10 18:33:22 [INFO]: Processing frame 560 (12.90 fps)
2021-11-10 18:33:24 [INFO]: Processing frame 580 (12.90 fps)
2021-11-10 18:33:25 [INFO]: Processing frame 600 (12.90 fps)
2021-11-10 18:33:27 [INFO]: Processing frame 620 (12.91 fps)
2021-11-10 18:33:29 [INFO]: Processing frame 640 (12.91 fps)
2021-11-10 18:33:31 [INFO]: Processing frame 660 (12.91 fps)
2021-11-10 18:33:32 [INFO]: Processing frame 680 (12.91 fps)
2021-11-10 18:33:34 [INFO]: Processing frame 700 (12.91 fps)
2021-11-10 18:33:36 [INFO]: Processing frame 720 (12.90 fps)
2021-11-10 18:33:37 [INFO]: Processing frame 740 (12.90 fps)
2021-11-10 18:33:39 [INFO]: Processing frame 760 (12.90 fps)
2021-11-10 18:33:41 [INFO]: Processing frame 780 (12.90 fps)
2021-11-10 18:33:43 [INFO]: Processing frame 800 (12.90 fps)
2021-11-10 18:33:44 [INFO]: Processing frame 820 (12.90 fps)
2021-11-10 18:33:46 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-05-SDP.txt
2021-11-10 18:33:46 [INFO]: Evaluate seq: MOT17-05-SDP
2021-11-10 18:33:46 [INFO]: start seq: MOT17-09-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:33:47 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:33:49 [INFO]: Processing frame 20 (12.87 fps)
2021-11-10 18:33:51 [INFO]: Processing frame 40 (12.79 fps)
2021-11-10 18:33:53 [INFO]: Processing frame 60 (12.81 fps)
2021-11-10 18:33:56 [INFO]: Processing frame 80 (12.83 fps)
2021-11-10 18:33:58 [INFO]: Processing frame 100 (12.85 fps)
2021-11-10 18:34:00 [INFO]: Processing frame 120 (12.87 fps)
2021-11-10 18:34:02 [INFO]: Processing frame 140 (12.90 fps)
2021-11-10 18:34:04 [INFO]: Processing frame 160 (12.91 fps)
2021-11-10 18:34:06 [INFO]: Processing frame 180 (12.91 fps)
2021-11-10 18:34:09 [INFO]: Processing frame 200 (12.90 fps)
2021-11-10 18:34:11 [INFO]: Processing frame 220 (12.90 fps)
2021-11-10 18:34:13 [INFO]: Processing frame 240 (12.89 fps)
2021-11-10 18:34:15 [INFO]: Processing frame 260 (12.88 fps)
2021-11-10 18:34:17 [INFO]: Processing frame 280 (12.88 fps)
2021-11-10 18:34:20 [INFO]: Processing frame 300 (12.88 fps)
2021-11-10 18:34:22 [INFO]: Processing frame 320 (12.88 fps)
2021-11-10 18:34:24 [INFO]: Processing frame 340 (12.89 fps)
2021-11-10 18:34:26 [INFO]: Processing frame 360 (12.89 fps)
2021-11-10 18:34:28 [INFO]: Processing frame 380 (12.88 fps)
2021-11-10 18:34:31 [INFO]: Processing frame 400 (12.87 fps)
2021-11-10 18:34:33 [INFO]: Processing frame 420 (12.87 fps)
2021-11-10 18:34:35 [INFO]: Processing frame 440 (12.86 fps)
2021-11-10 18:34:37 [INFO]: Processing frame 460 (12.85 fps)
2021-11-10 18:34:40 [INFO]: Processing frame 480 (12.84 fps)
2021-11-10 18:34:42 [INFO]: Processing frame 500 (12.84 fps)
2021-11-10 18:34:44 [INFO]: Processing frame 520 (12.84 fps)
2021-11-10 18:34:45 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-09-SDP.txt
2021-11-10 18:34:45 [INFO]: Evaluate seq: MOT17-09-SDP
2021-11-10 18:34:45 [INFO]: start seq: MOT17-10-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:34:45 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:34:47 [INFO]: Processing frame 20 (12.57 fps)
2021-11-10 18:34:50 [INFO]: Processing frame 40 (12.53 fps)
2021-11-10 18:34:52 [INFO]: Processing frame 60 (12.49 fps)
2021-11-10 18:34:54 [INFO]: Processing frame 80 (12.52 fps)
2021-11-10 18:34:56 [INFO]: Processing frame 100 (12.55 fps)
2021-11-10 18:34:58 [INFO]: Processing frame 120 (12.57 fps)
2021-11-10 18:35:00 [INFO]: Processing frame 140 (12.58 fps)
2021-11-10 18:35:03 [INFO]: Processing frame 160 (12.59 fps)
2021-11-10 18:35:05 [INFO]: Processing frame 180 (12.59 fps)
2021-11-10 18:35:07 [INFO]: Processing frame 200 (12.59 fps)
2021-11-10 18:35:09 [INFO]: Processing frame 220 (12.58 fps)
2021-11-10 18:35:11 [INFO]: Processing frame 240 (12.58 fps)
2021-11-10 18:35:14 [INFO]: Processing frame 260 (12.59 fps)
2021-11-10 18:35:16 [INFO]: Processing frame 280 (12.59 fps)
2021-11-10 18:35:18 [INFO]: Processing frame 300 (12.58 fps)
2021-11-10 18:35:20 [INFO]: Processing frame 320 (12.58 fps)
2021-11-10 18:35:22 [INFO]: Processing frame 340 (12.58 fps)
2021-11-10 18:35:24 [INFO]: Processing frame 360 (12.58 fps)
2021-11-10 18:35:26 [INFO]: Processing frame 380 (12.59 fps)
2021-11-10 18:35:29 [INFO]: Processing frame 400 (12.60 fps)
2021-11-10 18:35:31 [INFO]: Processing frame 420 (12.61 fps)
2021-11-10 18:35:33 [INFO]: Processing frame 440 (12.61 fps)
2021-11-10 18:35:35 [INFO]: Processing frame 460 (12.62 fps)
2021-11-10 18:35:37 [INFO]: Processing frame 480 (12.62 fps)
2021-11-10 18:35:39 [INFO]: Processing frame 500 (12.62 fps)
2021-11-10 18:35:42 [INFO]: Processing frame 520 (12.61 fps)
2021-11-10 18:35:44 [INFO]: Processing frame 540 (12.62 fps)
2021-11-10 18:35:46 [INFO]: Processing frame 560 (12.61 fps)
2021-11-10 18:35:48 [INFO]: Processing frame 580 (12.61 fps)
2021-11-10 18:35:50 [INFO]: Processing frame 600 (12.61 fps)
2021-11-10 18:35:52 [INFO]: Processing frame 620 (12.61 fps)
2021-11-10 18:35:55 [INFO]: Processing frame 640 (12.61 fps)
2021-11-10 18:35:56 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-10-SDP.txt
2021-11-10 18:35:56 [INFO]: Evaluate seq: MOT17-10-SDP
2021-11-10 18:35:57 [INFO]: start seq: MOT17-11-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:35:57 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:35:59 [INFO]: Processing frame 20 (12.69 fps)
2021-11-10 18:36:01 [INFO]: Processing frame 40 (12.61 fps)
2021-11-10 18:36:04 [INFO]: Processing frame 60 (12.62 fps)
2021-11-10 18:36:06 [INFO]: Processing frame 80 (12.64 fps)
2021-11-10 18:36:08 [INFO]: Processing frame 100 (12.64 fps)
2021-11-10 18:36:10 [INFO]: Processing frame 120 (12.65 fps)
2021-11-10 18:36:12 [INFO]: Processing frame 140 (12.66 fps)
2021-11-10 18:36:15 [INFO]: Processing frame 160 (12.66 fps)
2021-11-10 18:36:17 [INFO]: Processing frame 180 (12.67 fps)
2021-11-10 18:36:19 [INFO]: Processing frame 200 (12.67 fps)
2021-11-10 18:36:21 [INFO]: Processing frame 220 (12.67 fps)
2021-11-10 18:36:24 [INFO]: Processing frame 240 (12.67 fps)
2021-11-10 18:36:26 [INFO]: Processing frame 260 (12.68 fps)
2021-11-10 18:36:28 [INFO]: Processing frame 280 (12.67 fps)
2021-11-10 18:36:30 [INFO]: Processing frame 300 (12.68 fps)
2021-11-10 18:36:32 [INFO]: Processing frame 320 (12.67 fps)
2021-11-10 18:36:35 [INFO]: Processing frame 340 (12.68 fps)
2021-11-10 18:36:37 [INFO]: Processing frame 360 (12.70 fps)
2021-11-10 18:36:39 [INFO]: Processing frame 380 (12.71 fps)
2021-11-10 18:36:41 [INFO]: Processing frame 400 (12.72 fps)
2021-11-10 18:36:43 [INFO]: Processing frame 420 (12.73 fps)
2021-11-10 18:36:46 [INFO]: Processing frame 440 (12.73 fps)
2021-11-10 18:36:48 [INFO]: Processing frame 460 (12.72 fps)
2021-11-10 18:36:50 [INFO]: Processing frame 480 (12.73 fps)
2021-11-10 18:36:52 [INFO]: Processing frame 500 (12.73 fps)
2021-11-10 18:36:54 [INFO]: Processing frame 520 (12.73 fps)
2021-11-10 18:36:57 [INFO]: Processing frame 540 (12.73 fps)
2021-11-10 18:36:59 [INFO]: Processing frame 560 (12.73 fps)
2021-11-10 18:37:01 [INFO]: Processing frame 580 (12.73 fps)
2021-11-10 18:37:03 [INFO]: Processing frame 600 (12.73 fps)
2021-11-10 18:37:06 [INFO]: Processing frame 620 (12.73 fps)
2021-11-10 18:37:08 [INFO]: Processing frame 640 (12.74 fps)
2021-11-10 18:37:10 [INFO]: Processing frame 660 (12.74 fps)
2021-11-10 18:37:12 [INFO]: Processing frame 680 (12.75 fps)
2021-11-10 18:37:14 [INFO]: Processing frame 700 (12.75 fps)
2021-11-10 18:37:16 [INFO]: Processing frame 720 (12.75 fps)
2021-11-10 18:37:19 [INFO]: Processing frame 740 (12.76 fps)
2021-11-10 18:37:21 [INFO]: Processing frame 760 (12.76 fps)
2021-11-10 18:37:23 [INFO]: Processing frame 780 (12.76 fps)
2021-11-10 18:37:25 [INFO]: Processing frame 800 (12.76 fps)
2021-11-10 18:37:28 [INFO]: Processing frame 820 (12.76 fps)
2021-11-10 18:37:30 [INFO]: Processing frame 840 (12.76 fps)
2021-11-10 18:37:32 [INFO]: Processing frame 860 (12.75 fps)
2021-11-10 18:37:34 [INFO]: Processing frame 880 (12.75 fps)
2021-11-10 18:37:37 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-11-SDP.txt
2021-11-10 18:37:37 [INFO]: Evaluate seq: MOT17-11-SDP
2021-11-10 18:37:37 [INFO]: start seq: MOT17-13-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:37:37 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:37:40 [INFO]: Processing frame 20 (12.29 fps)
2021-11-10 18:37:42 [INFO]: Processing frame 40 (12.28 fps)
2021-11-10 18:37:45 [INFO]: Processing frame 60 (12.05 fps)
2021-11-10 18:37:47 [INFO]: Processing frame 80 (12.02 fps)
2021-11-10 18:37:49 [INFO]: Processing frame 100 (11.98 fps)
2021-11-10 18:37:52 [INFO]: Processing frame 120 (11.98 fps)
2021-11-10 18:37:54 [INFO]: Processing frame 140 (11.97 fps)
2021-11-10 18:37:56 [INFO]: Processing frame 160 (12.01 fps)
2021-11-10 18:37:59 [INFO]: Processing frame 180 (12.06 fps)
2021-11-10 18:38:01 [INFO]: Processing frame 200 (12.12 fps)
2021-11-10 18:38:03 [INFO]: Processing frame 220 (12.15 fps)
2021-11-10 18:38:05 [INFO]: Processing frame 240 (12.18 fps)
2021-11-10 18:38:08 [INFO]: Processing frame 260 (12.22 fps)
2021-11-10 18:38:10 [INFO]: Processing frame 280 (12.25 fps)
2021-11-10 18:38:12 [INFO]: Processing frame 300 (12.27 fps)
2021-11-10 18:38:14 [INFO]: Processing frame 320 (12.29 fps)
2021-11-10 18:38:17 [INFO]: Processing frame 340 (12.30 fps)
2021-11-10 18:38:19 [INFO]: Processing frame 360 (12.31 fps)
2021-11-10 18:38:21 [INFO]: Processing frame 380 (12.31 fps)
2021-11-10 18:38:24 [INFO]: Processing frame 400 (12.33 fps)
2021-11-10 18:38:26 [INFO]: Processing frame 420 (12.34 fps)
2021-11-10 18:38:28 [INFO]: Processing frame 440 (12.36 fps)
2021-11-10 18:38:30 [INFO]: Processing frame 460 (12.36 fps)
2021-11-10 18:38:32 [INFO]: Processing frame 480 (12.38 fps)
2021-11-10 18:38:35 [INFO]: Processing frame 500 (12.40 fps)
2021-11-10 18:38:37 [INFO]: Processing frame 520 (12.42 fps)
2021-11-10 18:38:39 [INFO]: Processing frame 540 (12.44 fps)
2021-11-10 18:38:41 [INFO]: Processing frame 560 (12.46 fps)
2021-11-10 18:38:43 [INFO]: Processing frame 580 (12.48 fps)
2021-11-10 18:38:45 [INFO]: Processing frame 600 (12.50 fps)
2021-11-10 18:38:48 [INFO]: Processing frame 620 (12.51 fps)
2021-11-10 18:38:50 [INFO]: Processing frame 640 (12.53 fps)
2021-11-10 18:38:52 [INFO]: Processing frame 660 (12.54 fps)
2021-11-10 18:38:54 [INFO]: Processing frame 680 (12.56 fps)
2021-11-10 18:38:56 [INFO]: Processing frame 700 (12.57 fps)
2021-11-10 18:38:58 [INFO]: Processing frame 720 (12.59 fps)
2021-11-10 18:39:01 [INFO]: Processing frame 740 (12.60 fps)
2021-11-10 18:39:02 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-13-SDP.txt
2021-11-10 18:39:02 [INFO]: Evaluate seq: MOT17-13-SDP
2021-11-10 18:39:02 [INFO]: Time elapsed: 423.51 seconds, FPS: 12.55
IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm
MOT17-02-SDP 53.0% 65.9% 44.4% 62.9% 93.4% 62 16 35 11 824 6887 219 584 57.3% 0.209 116 70 14
MOT17-04-SDP 82.0% 85.5% 78.8% 89.6% 97.2% 83 68 11 4 1242 4934 143 557 86.7% 0.184 34 62 5
MOT17-05-SDP 68.0% 80.7% 58.8% 69.5% 95.4% 133 33 79 21 234 2112 93 215 64.7% 0.181 63 48 31
MOT17-09-SDP 68.2% 77.1% 61.1% 78.0% 98.4% 26 15 11 0 68 1173 46 87 75.8% 0.171 21 23 5
MOT17-10-SDP 64.9% 75.0% 57.1% 72.0% 94.6% 57 22 34 1 529 3595 128 528 66.9% 0.233 61 48 4
MOT17-11-SDP 78.2% 85.4% 72.1% 81.3% 96.4% 75 38 28 9 287 1761 61 145 77.6% 0.157 26 29 8
MOT17-13-SDP 70.2% 79.8% 62.6% 73.0% 93.1% 110 43 55 12 634 3138 120 468 66.5% 0.234 66 55 28
OVERALL 72.7% 80.5% 66.3% 79.0% 95.9% 546 235 253 58 3818 23600 810 2584 74.9% 0.194 387 335 95
(FairMOT) caozilong@caozilong-RedmiBook-14:~/FairMOT/src$
下图是执行结束后输出的运行数据:
虽然MX250显卡是很挫的,但是从运行来看,还是对计算过程加持不少。
结束后,在~/dataset/MOT17/images/results/目录下输出针对每个目录的推理结果文件,如下图:
我们随便打开一个,比如MOT17-02-SDP.txt,上面的数据是检测到的物体位置和门类信息,推测第一列的数字达标图像序号,根据这个结果,说明第一章图像中找到了19个物体。
同时,生成的excel表格文件summary_MOT17_test_public_dla34.xlsx中包含了每个样本的统计信息:
这里有一个疑问,就是似乎以上过程是一个推理的过程,但是用的却是train目录的数据集,关于数据集很好验证,因为使用的MOT17-02-SDP,MOT17-04-SDP,MOT17-05-SDP,MOT17-09-SDP,MOT17-10-SDP,MOT17-11-SDP,MOT17-13-SDP都是TRAIN目录的数据集,现在对模型不是很熟悉,以后在寻找答案吧,或者那位看到帖子的朋友恰好知道答案,为我解答一下。
可视化输出结果:
以上方式产生的结果并非可视化的,浏览track.py代码,发现对main的调用有可操作的空间:
我们将show_image,save_images,save_videos全部修改为True看一下,修改后整个工作空间的改动情况是这个样子:
修改后,再次执行命令:
python track.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
这次正常运行的输出如下:
(FairMOT) caozilong@caozilong-RedmiBook-14:~/FairMOT/src$ python track.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
Fix size testing.
training chunk_sizes: [6, 6]
The output will be saved to /home/caozilong/FairMOT/src/lib/../../exp/mot/default
heads {'hm': 1, 'wh': 4, 'id': 64, 'reg': 2}
2021-11-10 18:58:02 [INFO]: start seq: MOT17-02-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:58:04 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:58:07 [INFO]: Processing frame 20 (11.10 fps)
2021-11-10 18:58:10 [INFO]: Processing frame 40 (11.90 fps)
2021-11-10 18:58:12 [INFO]: Processing frame 60 (12.16 fps)
2021-11-10 18:58:15 [INFO]: Processing frame 80 (12.32 fps)
2021-11-10 18:58:18 [INFO]: Processing frame 100 (12.41 fps)
2021-11-10 18:58:21 [INFO]: Processing frame 120 (12.47 fps)
2021-11-10 18:58:23 [INFO]: Processing frame 140 (12.52 fps)
2021-11-10 18:58:26 [INFO]: Processing frame 160 (12.54 fps)
2021-11-10 18:58:29 [INFO]: Processing frame 180 (12.56 fps)
2021-11-10 18:58:32 [INFO]: Processing frame 200 (12.58 fps)
2021-11-10 18:58:34 [INFO]: Processing frame 220 (12.59 fps)
2021-11-10 18:58:37 [INFO]: Processing frame 240 (12.60 fps)
2021-11-10 18:58:40 [INFO]: Processing frame 260 (12.61 fps)
2021-11-10 18:58:43 [INFO]: Processing frame 280 (12.60 fps)
2021-11-10 18:58:46 [INFO]: Processing frame 300 (12.58 fps)
2021-11-10 18:58:49 [INFO]: Processing frame 320 (12.56 fps)
2021-11-10 18:58:52 [INFO]: Processing frame 340 (12.52 fps)
2021-11-10 18:58:55 [INFO]: Processing frame 360 (12.50 fps)
2021-11-10 18:58:58 [INFO]: Processing frame 380 (12.46 fps)
2021-11-10 18:59:01 [INFO]: Processing frame 400 (12.45 fps)
2021-11-10 18:59:04 [INFO]: Processing frame 420 (12.43 fps)
2021-11-10 18:59:07 [INFO]: Processing frame 440 (12.44 fps)
2021-11-10 18:59:10 [INFO]: Processing frame 460 (12.45 fps)
2021-11-10 18:59:13 [INFO]: Processing frame 480 (12.45 fps)
2021-11-10 18:59:16 [INFO]: Processing frame 500 (12.45 fps)
2021-11-10 18:59:19 [INFO]: Processing frame 520 (12.45 fps)
2021-11-10 18:59:22 [INFO]: Processing frame 540 (12.46 fps)
2021-11-10 18:59:24 [INFO]: Processing frame 560 (12.45 fps)
2021-11-10 18:59:27 [INFO]: Processing frame 580 (12.45 fps)
2021-11-10 18:59:30 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-02-SDP.txt
2021-11-10 18:59:30 [INFO]: Evaluate seq: MOT17-02-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-02-SDP/%05d.jpg':
Duration: 00:00:24.00, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-02-SDP/MOT17-02-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 600 fps=0.0 q=-1.0 Lsize= 199275kB time=00:00:23.96 bitrate=68132.6kbits/s speed=80.7x
video:199271kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.002084%
2021-11-10 18:59:32 [INFO]: start seq: MOT17-04-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 18:59:32 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 18:59:35 [INFO]: Processing frame 20 (12.15 fps)
2021-11-10 18:59:38 [INFO]: Processing frame 40 (12.09 fps)
2021-11-10 18:59:42 [INFO]: Processing frame 60 (12.02 fps)
2021-11-10 18:59:45 [INFO]: Processing frame 80 (12.03 fps)
2021-11-10 18:59:48 [INFO]: Processing frame 100 (12.04 fps)
2021-11-10 18:59:51 [INFO]: Processing frame 120 (12.05 fps)
2021-11-10 18:59:54 [INFO]: Processing frame 140 (12.06 fps)
2021-11-10 18:59:57 [INFO]: Processing frame 160 (12.05 fps)
2021-11-10 19:00:00 [INFO]: Processing frame 180 (12.06 fps)
2021-11-10 19:00:03 [INFO]: Processing frame 200 (12.06 fps)
2021-11-10 19:00:06 [INFO]: Processing frame 220 (12.07 fps)
2021-11-10 19:00:09 [INFO]: Processing frame 240 (12.06 fps)
2021-11-10 19:00:12 [INFO]: Processing frame 260 (12.05 fps)
2021-11-10 19:00:15 [INFO]: Processing frame 280 (12.05 fps)
2021-11-10 19:00:18 [INFO]: Processing frame 300 (12.04 fps)
2021-11-10 19:00:21 [INFO]: Processing frame 320 (12.03 fps)
2021-11-10 19:00:24 [INFO]: Processing frame 340 (12.02 fps)
2021-11-10 19:00:28 [INFO]: Processing frame 360 (12.00 fps)
2021-11-10 19:00:31 [INFO]: Processing frame 380 (11.96 fps)
2021-11-10 19:00:34 [INFO]: Processing frame 400 (11.94 fps)
2021-11-10 19:00:37 [INFO]: Processing frame 420 (11.94 fps)
2021-11-10 19:00:40 [INFO]: Processing frame 440 (11.95 fps)
2021-11-10 19:00:43 [INFO]: Processing frame 460 (11.95 fps)
2021-11-10 19:00:46 [INFO]: Processing frame 480 (11.95 fps)
2021-11-10 19:00:49 [INFO]: Processing frame 500 (11.94 fps)
2021-11-10 19:00:52 [INFO]: Processing frame 520 (11.94 fps)
2021-11-10 19:00:56 [INFO]: Processing frame 540 (11.93 fps)
2021-11-10 19:00:59 [INFO]: Processing frame 560 (11.92 fps)
2021-11-10 19:01:02 [INFO]: Processing frame 580 (11.91 fps)
2021-11-10 19:01:05 [INFO]: Processing frame 600 (11.92 fps)
2021-11-10 19:01:08 [INFO]: Processing frame 620 (11.92 fps)
2021-11-10 19:01:11 [INFO]: Processing frame 640 (11.93 fps)
2021-11-10 19:01:14 [INFO]: Processing frame 660 (11.93 fps)
2021-11-10 19:01:17 [INFO]: Processing frame 680 (11.94 fps)
2021-11-10 19:01:20 [INFO]: Processing frame 700 (11.94 fps)
2021-11-10 19:01:24 [INFO]: Processing frame 720 (11.95 fps)
2021-11-10 19:01:27 [INFO]: Processing frame 740 (11.95 fps)
2021-11-10 19:01:30 [INFO]: Processing frame 760 (11.96 fps)
2021-11-10 19:01:33 [INFO]: Processing frame 780 (11.96 fps)
2021-11-10 19:01:36 [INFO]: Processing frame 800 (11.96 fps)
2021-11-10 19:01:39 [INFO]: Processing frame 820 (11.96 fps)
2021-11-10 19:01:42 [INFO]: Processing frame 840 (11.96 fps)
2021-11-10 19:01:45 [INFO]: Processing frame 860 (11.96 fps)
2021-11-10 19:01:48 [INFO]: Processing frame 880 (11.96 fps)
2021-11-10 19:01:51 [INFO]: Processing frame 900 (11.96 fps)
2021-11-10 19:01:54 [INFO]: Processing frame 920 (11.96 fps)
2021-11-10 19:01:57 [INFO]: Processing frame 940 (11.96 fps)
2021-11-10 19:02:00 [INFO]: Processing frame 960 (11.96 fps)
2021-11-10 19:02:03 [INFO]: Processing frame 980 (11.96 fps)
2021-11-10 19:02:06 [INFO]: Processing frame 1000 (11.95 fps)
2021-11-10 19:02:09 [INFO]: Processing frame 1020 (11.95 fps)
2021-11-10 19:02:13 [INFO]: Processing frame 1040 (11.95 fps)
2021-11-10 19:02:14 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-04-SDP.txt
2021-11-10 19:02:14 [INFO]: Evaluate seq: MOT17-04-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-04-SDP/%05d.jpg':
Duration: 00:00:42.00, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-04-SDP/MOT17-04-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 1050 fps=974 q=-1.0 Lsize= 444184kB time=00:00:41.96 bitrate=86719.5kbits/s speed=38.9x
video:444177kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.001561%
2021-11-10 19:02:18 [INFO]: start seq: MOT17-05-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 19:02:19 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 19:02:22 [INFO]: Processing frame 20 (12.89 fps)
2021-11-10 19:02:24 [INFO]: Processing frame 40 (12.95 fps)
2021-11-10 19:02:26 [INFO]: Processing frame 60 (12.99 fps)
2021-11-10 19:02:28 [INFO]: Processing frame 80 (13.02 fps)
2021-11-10 19:02:30 [INFO]: Processing frame 100 (12.97 fps)
2021-11-10 19:02:32 [INFO]: Processing frame 120 (12.99 fps)
2021-11-10 19:02:34 [INFO]: Processing frame 140 (12.97 fps)
2021-11-10 19:02:36 [INFO]: Processing frame 160 (12.97 fps)
2021-11-10 19:02:38 [INFO]: Processing frame 180 (12.98 fps)
2021-11-10 19:02:40 [INFO]: Processing frame 200 (12.99 fps)
2021-11-10 19:02:42 [INFO]: Processing frame 220 (12.99 fps)
2021-11-10 19:02:44 [INFO]: Processing frame 240 (12.99 fps)
2021-11-10 19:02:46 [INFO]: Processing frame 260 (12.97 fps)
2021-11-10 19:02:48 [INFO]: Processing frame 280 (12.97 fps)
2021-11-10 19:02:49 [INFO]: Processing frame 300 (12.97 fps)
2021-11-10 19:02:51 [INFO]: Processing frame 320 (12.97 fps)
2021-11-10 19:02:53 [INFO]: Processing frame 340 (12.97 fps)
2021-11-10 19:02:55 [INFO]: Processing frame 360 (12.97 fps)
2021-11-10 19:02:57 [INFO]: Processing frame 380 (12.98 fps)
2021-11-10 19:02:59 [INFO]: Processing frame 400 (12.98 fps)
2021-11-10 19:03:01 [INFO]: Processing frame 420 (12.98 fps)
2021-11-10 19:03:03 [INFO]: Processing frame 440 (12.98 fps)
2021-11-10 19:03:05 [INFO]: Processing frame 460 (12.96 fps)
2021-11-10 19:03:06 [INFO]: Processing frame 480 (12.97 fps)
2021-11-10 19:03:08 [INFO]: Processing frame 500 (12.97 fps)
2021-11-10 19:03:10 [INFO]: Processing frame 520 (12.97 fps)
2021-11-10 19:03:12 [INFO]: Processing frame 540 (12.97 fps)
2021-11-10 19:03:14 [INFO]: Processing frame 560 (12.97 fps)
2021-11-10 19:03:16 [INFO]: Processing frame 580 (12.97 fps)
2021-11-10 19:03:17 [INFO]: Processing frame 600 (12.97 fps)
2021-11-10 19:03:19 [INFO]: Processing frame 620 (12.98 fps)
2021-11-10 19:03:21 [INFO]: Processing frame 640 (12.98 fps)
2021-11-10 19:03:23 [INFO]: Processing frame 660 (12.98 fps)
2021-11-10 19:03:25 [INFO]: Processing frame 680 (12.98 fps)
2021-11-10 19:03:27 [INFO]: Processing frame 700 (12.98 fps)
2021-11-10 19:03:29 [INFO]: Processing frame 720 (12.97 fps)
2021-11-10 19:03:30 [INFO]: Processing frame 740 (12.97 fps)
2021-11-10 19:03:32 [INFO]: Processing frame 760 (12.97 fps)
2021-11-10 19:03:34 [INFO]: Processing frame 780 (12.96 fps)
2021-11-10 19:03:36 [INFO]: Processing frame 800 (12.96 fps)
2021-11-10 19:03:38 [INFO]: Processing frame 820 (12.96 fps)
2021-11-10 19:03:40 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-05-SDP.txt
2021-11-10 19:03:40 [INFO]: Evaluate seq: MOT17-05-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-05-SDP/%05d.jpg':
Duration: 00:00:33.48, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 640x480 [SAR 1:1 DAR 4:3], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-05-SDP/MOT17-05-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 640x480 [SAR 1:1 DAR 4:3], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 837 fps=0.0 q=-1.0 Lsize= 65108kB time=00:00:33.44 bitrate=15949.9kbits/s speed= 500x
video:65104kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.007148%
2021-11-10 19:03:41 [INFO]: start seq: MOT17-09-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 19:03:41 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 19:03:44 [INFO]: Processing frame 20 (12.93 fps)
2021-11-10 19:03:47 [INFO]: Processing frame 40 (13.00 fps)
2021-11-10 19:03:50 [INFO]: Processing frame 60 (12.97 fps)
2021-11-10 19:03:53 [INFO]: Processing frame 80 (12.91 fps)
2021-11-10 19:03:55 [INFO]: Processing frame 100 (12.93 fps)
2021-11-10 19:03:58 [INFO]: Processing frame 120 (12.94 fps)
2021-11-10 19:04:02 [INFO]: Processing frame 140 (12.94 fps)
2021-11-10 19:04:05 [INFO]: Processing frame 160 (12.93 fps)
2021-11-10 19:04:08 [INFO]: Processing frame 180 (12.93 fps)
2021-11-10 19:04:11 [INFO]: Processing frame 200 (12.91 fps)
2021-11-10 19:04:14 [INFO]: Processing frame 220 (12.89 fps)
2021-11-10 19:04:17 [INFO]: Processing frame 240 (12.88 fps)
2021-11-10 19:04:20 [INFO]: Processing frame 260 (12.89 fps)
2021-11-10 19:04:23 [INFO]: Processing frame 280 (12.89 fps)
2021-11-10 19:04:26 [INFO]: Processing frame 300 (12.89 fps)
2021-11-10 19:04:29 [INFO]: Processing frame 320 (12.88 fps)
2021-11-10 19:04:32 [INFO]: Processing frame 340 (12.88 fps)
2021-11-10 19:04:35 [INFO]: Processing frame 360 (12.89 fps)
2021-11-10 19:04:38 [INFO]: Processing frame 380 (12.89 fps)
2021-11-10 19:04:41 [INFO]: Processing frame 400 (12.89 fps)
2021-11-10 19:04:43 [INFO]: Processing frame 420 (12.89 fps)
2021-11-10 19:04:46 [INFO]: Processing frame 440 (12.90 fps)
2021-11-10 19:04:49 [INFO]: Processing frame 460 (12.90 fps)
2021-11-10 19:04:52 [INFO]: Processing frame 480 (12.90 fps)
2021-11-10 19:04:55 [INFO]: Processing frame 500 (12.91 fps)
2021-11-10 19:04:58 [INFO]: Processing frame 520 (12.91 fps)
2021-11-10 19:04:59 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-09-SDP.txt
2021-11-10 19:04:59 [INFO]: Evaluate seq: MOT17-09-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-09-SDP/%05d.jpg':
Duration: 00:00:21.00, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-09-SDP/MOT17-09-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 525 fps=0.0 q=-1.0 Lsize= 212776kB time=00:00:20.96 bitrate=83160.9kbits/s speed= 69x
video:212772kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.001820%
2021-11-10 19:05:00 [INFO]: start seq: MOT17-10-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 19:05:00 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 19:05:03 [INFO]: Processing frame 20 (12.59 fps)
2021-11-10 19:05:06 [INFO]: Processing frame 40 (12.57 fps)
2021-11-10 19:05:09 [INFO]: Processing frame 60 (12.54 fps)
2021-11-10 19:05:12 [INFO]: Processing frame 80 (12.56 fps)
2021-11-10 19:05:15 [INFO]: Processing frame 100 (12.58 fps)
2021-11-10 19:05:18 [INFO]: Processing frame 120 (12.60 fps)
2021-11-10 19:05:21 [INFO]: Processing frame 140 (12.60 fps)
2021-11-10 19:05:24 [INFO]: Processing frame 160 (12.60 fps)
2021-11-10 19:05:27 [INFO]: Processing frame 180 (12.60 fps)
2021-11-10 19:05:30 [INFO]: Processing frame 200 (12.59 fps)
2021-11-10 19:05:32 [INFO]: Processing frame 220 (12.59 fps)
2021-11-10 19:05:35 [INFO]: Processing frame 240 (12.58 fps)
2021-11-10 19:05:38 [INFO]: Processing frame 260 (12.57 fps)
2021-11-10 19:05:41 [INFO]: Processing frame 280 (12.57 fps)
2021-11-10 19:05:44 [INFO]: Processing frame 300 (12.56 fps)
2021-11-10 19:05:47 [INFO]: Processing frame 320 (12.55 fps)
2021-11-10 19:05:50 [INFO]: Processing frame 340 (12.55 fps)
2021-11-10 19:05:53 [INFO]: Processing frame 360 (12.55 fps)
2021-11-10 19:05:56 [INFO]: Processing frame 380 (12.56 fps)
2021-11-10 19:05:59 [INFO]: Processing frame 400 (12.56 fps)
2021-11-10 19:06:01 [INFO]: Processing frame 420 (12.57 fps)
2021-11-10 19:06:04 [INFO]: Processing frame 440 (12.57 fps)
2021-11-10 19:06:07 [INFO]: Processing frame 460 (12.58 fps)
2021-11-10 19:06:10 [INFO]: Processing frame 480 (12.58 fps)
2021-11-10 19:06:13 [INFO]: Processing frame 500 (12.58 fps)
2021-11-10 19:06:16 [INFO]: Processing frame 520 (12.59 fps)
2021-11-10 19:06:19 [INFO]: Processing frame 540 (12.59 fps)
2021-11-10 19:06:22 [INFO]: Processing frame 560 (12.59 fps)
2021-11-10 19:06:25 [INFO]: Processing frame 580 (12.59 fps)
2021-11-10 19:06:27 [INFO]: Processing frame 600 (12.59 fps)
2021-11-10 19:06:30 [INFO]: Processing frame 620 (12.58 fps)
2021-11-10 19:06:33 [INFO]: Processing frame 640 (12.59 fps)
2021-11-10 19:06:35 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-10-SDP.txt
2021-11-10 19:06:35 [INFO]: Evaluate seq: MOT17-10-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-10-SDP/%05d.jpg':
Duration: 00:00:26.16, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-10-SDP/MOT17-10-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 654 fps=0.0 q=-1.0 Lsize= 177800kB time=00:00:26.12 bitrate=55763.2kbits/s speed= 129x
video:177796kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.002547%
2021-11-10 19:06:36 [INFO]: start seq: MOT17-11-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 19:06:37 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 19:06:40 [INFO]: Processing frame 20 (12.61 fps)
2021-11-10 19:06:43 [INFO]: Processing frame 40 (12.66 fps)
2021-11-10 19:06:46 [INFO]: Processing frame 60 (12.68 fps)
2021-11-10 19:06:49 [INFO]: Processing frame 80 (12.72 fps)
2021-11-10 19:06:52 [INFO]: Processing frame 100 (12.73 fps)
2021-11-10 19:06:55 [INFO]: Processing frame 120 (12.73 fps)
2021-11-10 19:06:58 [INFO]: Processing frame 140 (12.73 fps)
2021-11-10 19:07:00 [INFO]: Processing frame 160 (12.75 fps)
2021-11-10 19:07:03 [INFO]: Processing frame 180 (12.76 fps)
2021-11-10 19:07:06 [INFO]: Processing frame 200 (12.78 fps)
2021-11-10 19:07:09 [INFO]: Processing frame 220 (12.79 fps)
2021-11-10 19:07:12 [INFO]: Processing frame 240 (12.81 fps)
2021-11-10 19:07:15 [INFO]: Processing frame 260 (12.81 fps)
2021-11-10 19:07:18 [INFO]: Processing frame 280 (12.81 fps)
2021-11-10 19:07:20 [INFO]: Processing frame 300 (12.82 fps)
2021-11-10 19:07:23 [INFO]: Processing frame 320 (12.82 fps)
2021-11-10 19:07:26 [INFO]: Processing frame 340 (12.82 fps)
2021-11-10 19:07:29 [INFO]: Processing frame 360 (12.83 fps)
2021-11-10 19:07:32 [INFO]: Processing frame 380 (12.83 fps)
2021-11-10 19:07:35 [INFO]: Processing frame 400 (12.84 fps)
2021-11-10 19:07:38 [INFO]: Processing frame 420 (12.84 fps)
2021-11-10 19:07:41 [INFO]: Processing frame 440 (12.84 fps)
2021-11-10 19:07:44 [INFO]: Processing frame 460 (12.84 fps)
2021-11-10 19:07:46 [INFO]: Processing frame 480 (12.84 fps)
2021-11-10 19:07:49 [INFO]: Processing frame 500 (12.84 fps)
2021-11-10 19:07:52 [INFO]: Processing frame 520 (12.85 fps)
2021-11-10 19:07:55 [INFO]: Processing frame 540 (12.85 fps)
2021-11-10 19:07:58 [INFO]: Processing frame 560 (12.85 fps)
2021-11-10 19:08:01 [INFO]: Processing frame 580 (12.85 fps)
2021-11-10 19:08:04 [INFO]: Processing frame 600 (12.85 fps)
2021-11-10 19:08:07 [INFO]: Processing frame 620 (12.85 fps)
2021-11-10 19:08:10 [INFO]: Processing frame 640 (12.85 fps)
2021-11-10 19:08:12 [INFO]: Processing frame 660 (12.85 fps)
2021-11-10 19:08:15 [INFO]: Processing frame 680 (12.85 fps)
2021-11-10 19:08:18 [INFO]: Processing frame 700 (12.85 fps)
2021-11-10 19:08:21 [INFO]: Processing frame 720 (12.85 fps)
2021-11-10 19:08:24 [INFO]: Processing frame 740 (12.86 fps)
2021-11-10 19:08:27 [INFO]: Processing frame 760 (12.86 fps)
2021-11-10 19:08:30 [INFO]: Processing frame 780 (12.86 fps)
2021-11-10 19:08:33 [INFO]: Processing frame 800 (12.86 fps)
2021-11-10 19:08:36 [INFO]: Processing frame 820 (12.85 fps)
2021-11-10 19:08:38 [INFO]: Processing frame 840 (12.85 fps)
2021-11-10 19:08:41 [INFO]: Processing frame 860 (12.85 fps)
2021-11-10 19:08:44 [INFO]: Processing frame 880 (12.85 fps)
2021-11-10 19:08:47 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-11-SDP.txt
2021-11-10 19:08:47 [INFO]: Evaluate seq: MOT17-11-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-11-SDP/%05d.jpg':
Duration: 00:00:36.00, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-11-SDP/MOT17-11-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 900 fps=0.0 q=-1.0 Lsize= 284244kB time=00:00:35.96 bitrate=64753.1kbits/s speed=69.3x
video:284238kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.002168%
2021-11-10 19:08:49 [INFO]: start seq: MOT17-13-SDP
Creating model...
/home/caozilong/FairMOT/src/lib/models/networks/config/yolov5s.yaml
loaded ../model/fairmot_lite.pth, epoch 30
2021-11-10 19:08:49 [INFO]: Processing frame 0 (100000.00 fps)
2021-11-10 19:08:53 [INFO]: Processing frame 20 (12.33 fps)
2021-11-10 19:08:56 [INFO]: Processing frame 40 (12.33 fps)
2021-11-10 19:08:59 [INFO]: Processing frame 60 (12.36 fps)
2021-11-10 19:09:02 [INFO]: Processing frame 80 (12.38 fps)
2021-11-10 19:09:05 [INFO]: Processing frame 100 (12.36 fps)
2021-11-10 19:09:08 [INFO]: Processing frame 120 (12.36 fps)
2021-11-10 19:09:11 [INFO]: Processing frame 140 (12.35 fps)
2021-11-10 19:09:14 [INFO]: Processing frame 160 (12.37 fps)
2021-11-10 19:09:17 [INFO]: Processing frame 180 (12.39 fps)
2021-11-10 19:09:20 [INFO]: Processing frame 200 (12.38 fps)
2021-11-10 19:09:24 [INFO]: Processing frame 220 (12.34 fps)
2021-11-10 19:09:27 [INFO]: Processing frame 240 (12.34 fps)
2021-11-10 19:09:30 [INFO]: Processing frame 260 (12.35 fps)
2021-11-10 19:09:33 [INFO]: Processing frame 280 (12.36 fps)
2021-11-10 19:09:36 [INFO]: Processing frame 300 (12.35 fps)
2021-11-10 19:09:40 [INFO]: Processing frame 320 (12.36 fps)
2021-11-10 19:09:43 [INFO]: Processing frame 340 (12.35 fps)
2021-11-10 19:09:46 [INFO]: Processing frame 360 (12.35 fps)
2021-11-10 19:09:49 [INFO]: Processing frame 380 (12.34 fps)
2021-11-10 19:09:52 [INFO]: Processing frame 400 (12.34 fps)
2021-11-10 19:09:55 [INFO]: Processing frame 420 (12.35 fps)
2021-11-10 19:09:58 [INFO]: Processing frame 440 (12.36 fps)
2021-11-10 19:10:01 [INFO]: Processing frame 460 (12.38 fps)
2021-11-10 19:10:04 [INFO]: Processing frame 480 (12.40 fps)
2021-11-10 19:10:07 [INFO]: Processing frame 500 (12.42 fps)
2021-11-10 19:10:10 [INFO]: Processing frame 520 (12.44 fps)
2021-11-10 19:10:13 [INFO]: Processing frame 540 (12.47 fps)
2021-11-10 19:10:16 [INFO]: Processing frame 560 (12.48 fps)
2021-11-10 19:10:19 [INFO]: Processing frame 580 (12.49 fps)
2021-11-10 19:10:22 [INFO]: Processing frame 600 (12.51 fps)
2021-11-10 19:10:25 [INFO]: Processing frame 620 (12.52 fps)
2021-11-10 19:10:28 [INFO]: Processing frame 640 (12.52 fps)
2021-11-10 19:10:31 [INFO]: Processing frame 660 (12.54 fps)
2021-11-10 19:10:34 [INFO]: Processing frame 680 (12.55 fps)
2021-11-10 19:10:37 [INFO]: Processing frame 700 (12.55 fps)
2021-11-10 19:10:40 [INFO]: Processing frame 720 (12.55 fps)
2021-11-10 19:10:43 [INFO]: Processing frame 740 (12.56 fps)
2021-11-10 19:10:45 [INFO]: save results to /home/caozilong/dataset/MOT17/images/train/../results/MOT17_test_public_dla34/MOT17-13-SDP.txt
2021-11-10 19:10:45 [INFO]: Evaluate seq: MOT17-13-SDP
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
libavutil 55. 78.100 / 55. 78.100
libavcodec 57.107.100 / 57.107.100
libavformat 57. 83.100 / 57. 83.100
libavdevice 57. 10.100 / 57. 10.100
libavfilter 6.107.100 / 6.107.100
libavresample 3. 7. 0 / 3. 7. 0
libswscale 4. 8.100 / 4. 8.100
libswresample 2. 9.100 / 2. 9.100
libpostproc 54. 7.100 / 54. 7.100
Input #0, image2, from '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-13-SDP/%05d.jpg':
Duration: 00:00:30.00, start: 0.000000, bitrate: N/A
Stream #0:0: Video: mjpeg, yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], 25 fps, 25 tbr, 25 tbn, 25 tbc
Output #0, mp4, to '/home/caozilong/dataset/MOT17/images/train/../outputs/MOT17_test_public_dla34/MOT17-13-SDP/MOT17-13-SDP.mp4':
Metadata:
encoder : Lavf57.83.100
Stream #0:0: Video: mjpeg (mp4v / 0x7634706D), yuvj420p(pc, bt470bg/unknown/unknown), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 25 fps, 25 tbr, 12800 tbn, 25 tbc
Stream mapping:
Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 750 fps=0.0 q=-1.0 Lsize= 283293kB time=00:00:29.96 bitrate=77460.9kbits/s speed=68.3x
video:283288kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.001848%
2021-11-10 19:10:46 [INFO]: Time elapsed: 423.16 seconds, FPS: 12.56
IDF1 IDP IDR Rcll Prcn GT MT PT ML FP FN IDs FM MOTA MOTP IDt IDa IDm
MOT17-02-SDP 53.0% 65.9% 44.4% 62.9% 93.4% 62 16 35 11 824 6887 219 584 57.3% 0.209 116 70 14
MOT17-04-SDP 82.0% 85.5% 78.8% 89.6% 97.2% 83 68 11 4 1242 4934 143 557 86.7% 0.184 34 62 5
MOT17-05-SDP 68.0% 80.7% 58.8% 69.5% 95.4% 133 33 79 21 234 2112 93 215 64.7% 0.181 63 48 31
MOT17-09-SDP 68.2% 77.1% 61.1% 78.0% 98.4% 26 15 11 0 68 1173 46 87 75.8% 0.171 21 23 5
MOT17-10-SDP 64.9% 75.0% 57.1% 72.0% 94.6% 57 22 34 1 529 3595 128 528 66.9% 0.233 61 48 4
MOT17-11-SDP 78.2% 85.4% 72.1% 81.3% 96.4% 75 38 28 9 287 1761 61 145 77.6% 0.157 26 29 8
MOT17-13-SDP 70.2% 79.8% 62.6% 73.0% 93.1% 110 43 55 12 634 3138 120 468 66.5% 0.234 66 55 28
OVERALL 72.7% 80.5% 66.3% 79.0% 95.9% 546 235 253 58 3818 23600 810 2584 74.9% 0.194 387 335 95
(FairMOT) caozilong@caozilong-RedmiBook-14:~/FairMOT/src$
与上次输出相比,这里的输出多了FFMPEG的调用,以MOT17-13-SDP目录数据集为例,可以看到,在推理完成后,将推理图像转换成了MP4视频格式。
验证结果,这次的输出多创建了一个outputs目录,目录中以数据集命名,和上次相比,results目录的内容似乎没有什么变化。
以images/outputs/MOT17_test_public_dla34/MOT17-02-SDP目录为例,它的内容如下:
与Train目录中的MOT17-02-SDP相比,推理结果目录中文件名序号从0开始而非从1开始,并且它多出来一个视频文件。
推理结果中的图片已经被绘了框,它是对上一步骤操作中生成的TXT文本输出文件的可视化处理,对比同一张图:
查看视频文件,可以看到动态画框的效果:
至此,环境搭建完毕。
模型转换:
修改multitracker.py文件,添加转换为onnx模型的代码逻辑:
之后再次运行用例,结束后在SRC目录下生成了fairmot.onnx文件,中间循环执行了多次,每次生成的都会覆盖之前的,但数据内容没有变化,fairmot.onnx.bak就是对比用的临时拷贝。
用netron工具对生成的ONNX文件进行查看:
首先是输入部分:
其次是输出部分:
可以看出输入部分和输出部分的网络节点名字都是由于修改的代码决定的:
运行demo.py用例:
执行命令:
python demo.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64
输出LOG:
输出目录~/FairMOT/demos:
frame目录为多目标追踪输出带框图片:
多目标追踪的视频文件:
使用CPU完成推理:
为了加快推理速度,前面的操作使用了GPU的推理,为了有效利用GPU,我们安装了CUDA以及CUDANN,CUDABLAS等数学库,最后FPS可以达到12帧。其实,FairMOT环境也可以使用CPU推理,方式是在命令行中加入 --gpus=-1选项,这一点在代码中可以看出来:
FairMOT/src/lib/opts.py
完整命令
python demo.py mot --test_mot17 True --load_model ../model/fairmot_lite.pth --conf_thres 0.4 --arch yolo --reid_dim 64 --gpus=-1
关闭CUDA
再次执行:
YOLO
类似的事情YOLO也可以做,和FairMOT相比,YOLO多了识别,下图是在GPU上运行YOLOV3做目标检测和画框识别的例子。
与FairMOT的效果对比,YOLOV3除了多了识别之外,目标种类则支持的更多,至少,图中的白色汽车是FairMOT没有框出来的。
比较:
比较上面两幅图可以看到,YOLO的视频检测和FireMOT的视频检测似乎是类似的,都是在目标上绘制框,并且框都护跟随目标移动的。但是也有区别:
1。类别对MOT不太重要,所以MOT没有类别,而是用ID代替。
2。MOT有ID,并且在追踪过程中,ID保持不变,重点在“跟踪”。
下图中每个框左上角上的数字,就是ID。
FAIRMOT核心算法
FAIRMOT算法是一种目标跟踪算法,核心是目标检测+匈牙利算法+卡尔曼滤波算法,这也是目标跟踪类算法的通用实现套路。
目标检测算法用于提供目标物体的位置,卡尔曼滤波算法用于预测物体的下一个位置,因为自然移动的物体位置不会发生突变(大师兄的72变不属于自然移动的物体),所以短时间的遮挡是可以通过卡尔曼滤波算法进行预测的。无遮挡情况下的正行运动更不必提。当预测框和目标框的IOU分数达到一定的阈值,算法就认为追踪到了物体。匈牙利算法用于将TRACKER ID和目标筐之间进行绑定。匈牙利算法的流程,简单来说就是一个递归过程,尽可能找到让上一帧与当前帧目标一对一的匹配。
参考:
如果通过命令行的方式安装cuda,可以参考以下链接:
https://www.tensorflow.org/install/gpu