Jetson TX2 darknet安装与测试——wiznote笔记链接
目录
1 安装darknet
1-1 克隆darknet repo
git clone https://github.com/pjreddie/darknet.git
cd darknet
1-2 修改Makefile
# 备份Makefile
cp Makefile Makefile.bak
vim Makefile
由于Jetpack3.3已经安装好了CUDA、cuDNN和OpenCV,所以将Makefile里面的内容修改如下:
GPU=1
CUDNN=1
OPENCV=1
1-3 编译
make -j2
2 测试
2-1 测试OpenCV demo
./darknet imtest data/eagle.jpg
2-2 在一张测试图片上测试yolov3
# 下载yolov3-weights
wget https://pjreddie.com/media/files/yolov3.weights
# 测试
sudo ./darknet detect cfg/yolov3.cfg yolov3.weights data/dog.jpg
出现错误:76 Killed
这是因为yolov3网络结构比较庞大,导致内存不足,因此我们换一个比较小的网络yolov2来测试
2-3 在一张测试图片上测试yolov2
# 下载yolov2-weights
wget https://pjreddie.com/media/files/yolo.weights
# 测试
sudo ./darknet detect cfg/yolov2.cfg yolo.weights data/dog.jpg
输出结果:
mask_scale: Using default '1.000000'
Loading weights from yolo.weights...Done!
data/dog.jpg: Predicted in 0.410068 seconds.
dog: 82%
truck: 65%
bicycle: 85%
2-4 在一张测试图片上测试yolov3-tiny
# 下载yolov3-tiny.weights
wget https://pjreddie.com/media/files/yolov3-tiny.weights
# 测试
sudo ./darknet detect cfg/yolov3-tiny.cfg yolov3-tiny.weights data/dog.jpg
输出结果:
Loading weights from yolov3-tiny.weights...Done!
data/dog.jpg: Predicted in 0.178556 seconds.
dog: 57%
car: 51%
truck: 56%
car: 62%
bicycle: 59%
2-5 使用网络摄像头上测试yolov2
Jetson TX2自带的摄像头不知道什么原因不能使用,我们在USB集线器上外接一个USB摄像头,输入命令:
sudo ./darknet detector demo cfg/coco.data cfg/yolov2.cfg yolo.weights -c 1
检测帧率在2.5-3.5FPS之间
注意:这里一定要输入标识符-c 1
,否则会OpenCV会调用Jetson TX2默认的摄像头,导致出现错误:
mask_scale: Using default '1.000000'
Loding weights from yolo.weoghts...Done!
VIDEOIO ERROR: V4L2:Pixel format of incoming image is unsupported by OpenCV
Couldn't connect to webcam.
: Invalid argument
darknet: ./src/utils.c:256: error: Assertion '0' failed.
Aborted (core dymped)
2-6 使用网络摄像头测试yolov3-tiny
sudo ./darknet detector demo cfg/coco.data cfg/yolov3-tiny.cfg yolov3-tiny.weights -c 1
检测帧率在16-22FPS之间