• 博客(930)
  • 资源 (31)
  • 收藏
  • 关注

原创 Visualizing and Understanding Convolutional Networks

代码请见:https://github.com/amir-saniyan/ZFNet # Input: 224x224x3. with tf.name_scope('input'): self.X = tf.placeholder(dtype=tf.float32, shape=[N...

2019-05-29 01:30:32 145

原创 如何写CmakeList.txt文件使在qtcreator里面显示源文件

cmake_minimum_required(VERSION 3.5)FILE(GLOB incs include/pcl/gpu/kinfu/*.h*)FILE(GLOB srcs src/*.cpp src/*.h*)FILE(GLOB cuda src/cuda/*.cu src/cuda/*.h*)add_custom_target(whatever SOURCES ${incs}...

2018-05-22 19:28:15 1950

原创 git add submodule

first to the submodule you want to add, git init, add remote and pushgo back to the parent folder, git submodule add ./your_submodule

2022-05-31 16:49:45 374

原创 define macro in xcode project

TARGETS->Swift compiler - custom flags -> active compilation conditionsrefer to: https://thandang.medium.com/how-to-work-with-macro-preprocessor-in-xcode-project-1ca84d576411

2022-02-24 07:57:51 192

原创 manually download dataset and import in tensorflow

somtimes, you may have problem with tensorflow build-in downloading function like usingtfds.load(name="the300w_lp", split="train", with_info=True)This could be different reason. One way to workaround is to download the dataset manually as mentioned in

2021-10-23 05:34:16 146

转载 stop auto remove cuda

It depends on how you installed CUDA. If you ran sudo apt-get install , and that package was removed, then CUDA, an automatically installed package, becomes automatically removable. The simple fix is to mark the CUDA packages as manually installed:sudo .

2021-10-22 19:02:09 67

原创 keymap in ubuntu

inux中自启动目录是$XDG_CONFIG_DIRS/autostart和$XDG_CONFIG_HOME/autostart这两个,可以打印看一下,有的可能只有其中一个。create a file named 'swap.desktop` in the folder[Desktop Entry]Name=SwapExec=xmodmap -e "keycode 118 = Delete" && xmodmap -e "keycode 119 = Insert"Ter..

2021-10-18 17:10:13 79

原创 crop video in ffmpeg

ffmpeg -i 84.avi -filter:v "crop=1000:400:0:300" 84_crop.avi

2021-09-13 19:26:43 86

原创 stack videos in ffmpeg

ffmpeg -i 84.avi -i 100.avi -filter_complex "hstack,format=yuv420p" -c:v libx264 -crf 18 -preset veryfast h.mp4ffmpeg -i 84.avi -i 100.avi -filter_complex "vstack,format=yuv420p" -c:v libx264 -crf 18 -preset veryfast v.mp4

2021-09-13 18:42:51 37

原创 人像处理流程

temperature, tint, exposure, contrast, highlight, shadow, sharpening, noise reduction, hsl(调主色饱和)calibration, position, liquify( edge, body, eye)spot fixing( patching, healing brush, painting)色调, gradient map(用主体色调, 别用在皮肤, 主要背景)selective colo..

2021-09-06 05:28:58 95

原创 git ignore 某行

在项目开发中, 有些配置文件中的配置是因为机器会不一样的, 比较配置文件中的IP地址或数据库地址, 这时我们希望每个项目保持这些配置而不相互覆盖.这里我们希望在*.ini文件中所有以//#gitignore结尾的行保持不变首先我们在项目中建立一个.gitattributes文件, 里面写上*.ini filter=lineignore这句的意思在所有*.ini加上一个过滤器, 过滤器的名字叫lineignore然后我们建立一个.gitconfig文件, 里面写上git config --globa

2021-08-10 07:43:45 159

原创 loop one image for 10 secs

ffmpeg -loop 1 -i 4.png -c:v libx264 -t 10 2.mp4

2021-06-23 13:38:48 32

原创 ffmpeg create a video from 4th image and and slow down the play speed

ffmpeg -y -start_number 4 -i %d.png -filter:v “setpts=10*PTS” 1.avi

2021-06-23 13:30:09 40

转载 primal and dual linear problem


2021-05-15 00:22:09 85

原创 customize ubuntu theme color like pycharm grey

You can edit the gtk-main.css or nautilus.css files.I recommend you to try this under your ~/.themes and not system-wide. You may either change the background color of nautilus only or the base theme as a whole, which affects a lot more programs. In this

2021-03-08 03:58:01 167

原创 concatenate mutliple videos in ffmpeg

ffmpeg -i “concat:1.avi|2.avi|3.avi” -codec copy 123.avi

2021-03-04 21:19:35 80 1

原创 将远程文件的前10个复制到本地

normally, we use rsync like this:rsync -aP tj0012@access.eps.surrey.ac.uk:/vol/research/extol/data/BSLCP/Videos/Interview/ /home/seamanj/Workspace/Skeletor/data/videossh tj0012@access.eps.surrey.ac.uk 'find /vol/research/extol/tmp/Tao_BSLCP/videos_resiz

2021-02-20 07:49:48 56 1

原创 enable pen pressure in ps

check use windoes inkopen brush setting in ‘windows’ menuturn off the size jitter so that the pressure will not affect the brush sizein transfer, set the flow jitter to ‘pen pressure’ to allow pressure affectting the flow. we don’t want pressure affe.

2021-01-01 05:20:32 60

原创 clarity, texture and sharpening

clarity works on mid-tone, basically add the constrast to the mid-tone. you could rephrase it as “pouching” toolClarity is used to increase local contrast - which is the separation between tones on a smaller level.texture works on mid-frequency area,sh

2020-12-21 02:46:45 285

原创 自定义autograd function

在TSN代码中, segmentconsensus是一个自定义函数, 所以要写一下它对应的梯度运算# tj : https://blog.csdn.net/tsq292978891/article/details/79364140class SegmentConsensus(torch.autograd.Function): @staticmethod def forward(ctx, input_tensor, consensus_type, dim): ctx.

2020-11-24 15:31:35 269 2

转载 How to un-submodule a Git submodule?

https://stackoverflow.com/questions/1759587/how-to-un-submodule-a-git-submodulegit rm --cached submodule_path # delete reference to submodule HEAD (no trailing slash)git rm .gitmodules # if you have more than one submodules,

2020-11-18 04:15:19 43

原创 setup the printer on ubuntu for uniofsurrey

Download the Samsung driver packageUncompress the driver package and locate the ppd file# tar xvf SamsungPrinterInstaller.tar.gzThe “ppd” file is: ./uld/noarch/share/ppd/Samsung_X7600_Series.ppdInstall and configure SurreyPrintOpen http://localhost:63

2020-11-05 05:13:02 60

原创 how to connect to eduroam

2020-11-04 20:41:10 110

原创 check GPU mem size on condor

condor_status -compact -constraint 'TotalGPUs>0' -af:h machine TotalGPUs CUDADeviceName CUDACapability CUDADriverVersion CUDAGlobalMemoryMb CUDAComputeUnits CUDACoresPerCU HasNvidiaDriver

2020-11-02 17:45:37 91

原创 ubuntu下显示文件按数字排序

ls -v-v - Force unedited printing of non-graphic characters

2020-11-02 03:51:19 479

原创 DataParallel work on cpu

DataParallel work on GPU by default. so if you want load model on cpu, you may not want to use DataParallel when you save your model. However, if you load someone else’s pretrained model which saved with DataParallel, how should we load it on cpu?if we ju

2020-11-01 04:44:30 153

原创 some condor commands

condor server :new : ssh condorold : ssh cvssp-condor-matstercondor_userprio : check the user prioritycondor_status : check the machine status , for concise version, add -compact flagour priority...

2020-10-22 01:04:31 58

原创 ssh to surrey server

generate rsa keyon you laptop,ssh-keygen -t rsa -b 4096then you will get id_rsa(private key) and id_rsa.pub(public key)on your desktopcopy the public key into ~/.ssh/authorized_keysnow ever...

2020-10-21 07:15:50 50

原创 import python file in currently folder

when importing python file in currently folder, pycharm may not find the links. just right click on the folder and set as sources root, then pycharm can validate the links.

2020-10-20 17:16:26 133

原创 build openposewith opencv-2.4.13,cuda9(9.0 - 9.2)

基本的流程参见 docker image这里强调几点重点, 编辑opencv时, 要安装好gtk, 否则摄像头功能不能用, 要安装好libavcodec-dev libavformat-dev libavutil-dev libswscale-dev否则视频功能不能用安装完后, 用下面命令测试视频./build/examples/openpose/openpose.bin --video ./examples/media/video.avi #FROM ubuntu:18.04F

2020-10-16 15:08:36 80

原创 在wamp里面配置feehicms

https://github.com/liufee/feehicmshttpd.conf文件uncomment LoadModule rewrite_module modules/mod_rewrite.souncomment Include conf/extra/httpd-vhosts.confhttpd-vhosts.conf 文件<VirtualHost *:80> ServerAdmin seamanj@gmail.com DocumentRoot "C:/s

2020-10-03 20:15:45 104

原创 mount the remote disk to your local

cat /etc/auto.master## Sample auto.master file# This is a 'master' automounter map and it has the following format:# mount-point [map-type[,format]:]map [options]# For details of the format look...

2020-09-30 00:29:41 26

转载 From conda create requirements.txt for pip3

As the comment at the top indicates, the output ofconda list -e > requirements.txtcan be used to create a conda virtual environment withconda create --name --file requirements.txtbut this output isn’t in the right format for pip.If you want a file

2020-07-13 21:16:39 169

原创 qt UI design tips

当一个控件不能居中时, 尝试将其放入qwidget中去居中设计页面时, 先建立一个qFrame, 然后将所有控件放入qFrame中, 这样方便以后调整尺寸

2020-06-20 21:01:44 116

原创 no space left on device

check your project spacessh tj0012@access.eps.surrey.ac.ukssh cvpsys01(base) tj0012@cvpsys01:~$ check_quota /vol/research/SignPose Information for: SignPose ------------------------------------------------------------------ Quota: .

2020-06-20 00:20:22 33

转载 Renaming local and remote

Rename the local branch to the new namegit branch -m <old_name> <new_name>Delete the old branch on remote - where is, for example, origingit push --delete <old_name>Or shorter way to delete remote branch [:]git push :<old_name&

2020-06-19 16:22:04 114

原创 post processing photo tips

you may want to add contrast on mid-tone with bright-tone and dark-tone kept, then use lumenzia mid-tone selection and then adjust the contrast on it.if you want add contrast on certain area, say add contrast on water, there are two ways:a - lumenzi..

2020-06-18 06:21:00 119

原创 exposure blending

first, black layer mask on dark one, white brush the sky(over the horizontal line), lumizia selection, black brush the horizontal line. if there is white halo, exchange white and black brush several times.

2020-06-18 04:13:06 120

原创 PS 快捷键

x 切换前景背影色ctrl+j 复制当前选择图层ctrl+shift+alt+e 复制现有状态

2020-06-15 17:49:06 325

原创 copy files from remote to local

rsync -aP tj0012@access.eps.surrey.ac.uk:/vol/research/SignPose/tj/Workspace/Skeletor_New/Skeletor__step2_breakpoint_three_4_curves_val_whole/models/ /home/seamanj/Workspace/Skeletor/models

2020-06-14 12:39:39 64

average diffusion distance

average diffusion distance implemented in matlab


Farthest Sampling on 3D meshes with mesh kept based on diffusion distance

Farthest Sampling on 3D meshes with mesh kept based on diffusion distance implemented in matlab


Fast Marching on 3D Meshes with diffusion distance

Fast Marching on 3D Meshes with diffusion distance implemented in matlab


PCA implementation

PCA implementation in matlab


Farthest Sampling on 3D meshes with mesh kept

Farthest Sampling on 3D meshes with mesh kept


Fast Marching on 3D Meshes

implement fast marching on 3d meshes in matlab


farthest point sampling

implement farthest point sampling in matlab


2D Fast Marching Computations

implement 2d fast marching computations in matalb


princal component analysis implemented in matlab

princal component analysis implemented in matlab


as-conformal-as-possible surface registration

implement paper "as-conformal-as-possible surface registraion" in matlab


position based dynamics

position based dynamics implemented in matlab


mixed finite elements

implement mixed finite elements in matlab


laplacian interpolation

implement laplacian interpolation in matlab



solve QSqlDatabase: QMYSQL driver not loaded problem





2d perlin noise ocean

采用二维柏林噪声模拟的海洋 .zip


softmax regression

softmax regression 源代码 源码用的armadillo数学库,如何安装armadillo数学库,请看我的博客 http://blog.csdn.net/seamanj/article/details/49892715


shadow Mapping 阴影投影

shadow Mapping 阴影投影 主要介绍阴影投影实现的原理 和 实现方法



移位贴图(Displacement Mapping)的实现



投影贴图(projective texturing)的实现



立方体环境映射(cubic environment mapping)的实现,其中包括了反射,折射,菲涅尔效应,色散等知识.



3D 法线贴图(normal mapping)的实现



3D球形环境映射(3D Sphere environment mapping)的实现



球形环境映射(sphere environment mappint)原理的实现



USB监控器_去网页打开 使用到的工具,OD和W32DASM



DX 混合 纹理操作 包括多重纹理的操作,以及BACKBUFFER的读取



四元组的应用 包含: 1.从"按定轴旋转某角度"建立四元组 2.四元组的球形插值插值 3.四元组与3x3旋转矩阵的相互转换






矩阵各种求逆的方法 包括: 1.用矩阵行列式求逆 2.用高斯-约当消元法求逆 3.用LU分解求逆



矩阵各种求逆的方法 包括: 1.用矩阵行列式求逆 2.用高斯-约当消元法求逆 3.用LU分解求逆



四元组应用的实现 包括了 1.从"按指定轴旋转多少度"建立四元组 2.四元组的球形插值 3.四元组与3x3旋转矩阵的相互转换



TA创建的收藏夹 TA关注的收藏夹


取消 删除