程永强

Darknet - 分类模型 (.weights) 精度验证1. Validate The ModelNow we have to see how well our model is doing. We can calculate top-1 and top-2 validation accuracy using the valid command. We can run validation on a backup, the final weights file, or any saved epoch

Linux 获取文件名称生成列表 txt - create_filelist1. find/home/strong/MOTChallenge/MOT16/MOT16/train/MOT16-04/img1/ 文件夹下所有 *.jpg (000001.jpg - 001050.jpg) 的路径 + 图片名信息写入 txt 文件。1.1 llstrong@foreverstrong:~/MOTChallenge/MOT16/MOT16/train/MOT16-04/img1$ lltotal

Darknet - 检测模型 (.weights) 计算 mAP (mean average precision)1. mAP (mean average precision) or IoU (intersect over union)https://github.com/AlexeyAB/darknetIf training is stopped after 9000 iterations, to validate some of previous weights use this commands

cat - concatenate files and print on the standard output1. cat 文件合并 demo 1anchor.txtpositive.txtnegative.txtstrong@foreverstrong:~/TensorFlow/group_detection/simulation_data$ lltotal 32drwxrwxr-x 3 strong strong 4096 11月 21 15:03 ./d.

Validate Classification Model - Darknet Validation Command Line1. darknet_validation_command_line_classification.py#!/usr/bin/env python3# -*- coding: utf-8 -*-# Yongqiang Chengfrom __future__ import absolute_importfrom __future__ import divisionfr

darknet_AlexeyAB_191221 - person - p*****_yolov3P*****_YOLOv31. anchors2. p*****_yolov3.data/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-master/data/p*****_yolov3.dataclasses= 1train = /media/famu/DISK_DEEP/yongqiang/yolov3-tiny_prn/y

Eclipse - Darknet - Makefile project - Debug / Run - launch configurations

darknet - Tiny YOLOv3 test and training (测试 and 训练)https://pjreddie.com/darknet/yolo/https://github.com/AlexeyAB/darknet1. Tiny YOLOv3 - testWe have a very small model as well for constrained environments, yolov3-tiny. To use this model, first download

Darknet - When should I stop training? - 我什么时候应该停止训练？https://github.com/AlexeyAB/darknetUsually sufficient 2000 iterations for each class (object), but not less than 4000 iterations in total. But for a more precise definition when you should stop trainin

relie_activate & relie_gradient1. darknet - activations.h - relie_activatestatic inline float relie_activate(float x){ return (x > 0) ? x : .01 * x;}2. darknet - activations.h - relie_gradientstatic inline float relie_gradient(float

Darknet - How to improve object detection? - 如何提升物体检测？https://github.com/AlexeyAB/darknet1. before trainingset flag random=1 in your .cfg-file - it will increase precision by training YOLO for different resolutions.在 .cfg 文件中设置标志 random = 1，通过对 YOLO

构建 Darknet 分类器 (Tiny Darknet) 训练数据集 (color recognition 颜色识别/color classification 颜色分类)1. CompCars_Color9_Dataset2. parser## Created by foreverstrong on 12/31/2018#import osimport cv2color_classes = ["white", "gray", "black", "red", "green", "blu

构建 Darknet 分类器 (Tiny Darknet) 训练数据集 (front_rear recognition 车头车尾识别/front_rear classification 车头车尾分类)1. vehicle_source_data_20181015_autoparts2. parser## Created by foreverstrong on 12/31/2018#import osimport cv2# classification_classes = ["fron

Python + OpenCV 深度学习数据集批量规范为 .jpg1. jpg、jpeg、bmp 统一转换为 .jpg，形成规范统一数据集#!/usr/bin/env python3# -*- coding: utf-8 -*-# Yongqiang Chengfrom __future__ import absolute_importfrom __future__ import divisionfrom __future__ import print_functionimport

darknet_AlexeyAB_191221 - bag1. anchors2. p*****_yolov3_bag.data/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-master/data/p*****_yolov3_bag.dataclasses= 3train = /media/famu/DISK_DEEP/yongqiang/p*****_yolov3_bag/training_validation_test

Darknet - YOLO - /bin/sh: 1: nvcc: not found1. darket_workyongqiang@amax-server:~$ ssh yongqiang@192.168.1.3yongqiang@192.168.1.3's password: Welcome to Ubuntu 16.04.4 LTS (GNU/Linux 4.4.0-116-generic x86_64) * Documentation: https://help.ubuntu.

Darknet: Open Source Neural Networks in C - Classifying With Pre-Trained Modelshttps://pjreddie.com/darknet/Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation. You

darknet_AlexeyAB_191221 - person - yolov3-tinyYOLOv3-Tiny1. anchors2. yolov3-tiny.data/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-master/data/yolov3-tiny.dataclasses= 1train = /media/famu/DISK_DEEP/yongqiang/yolov3-tiny_prn/yolov3-ti

relu_activate &relu_gradient1. darknet - activations.h -relu_activatestatic inline float relu_activate(float x){ return x * (x > 0);}2. darknet - activations.h -relu_gradientstatic inline float relu_gradient(float x){ return ...

Darknet: Open Source Neural Networks in C - CIFAR-10 Dataset Mirrorhttps://pjreddie.com/projects/CIFAR-10 Dataset MirrorA mirror of the popular CIFAR-10 dataset, in png format.popular [ˈpɒpjələ(r)]：adj. 流行的，通俗的，受欢迎的，大众的，普及的1. CIFAR-10 Dataset Mirror

darknet_AlexeyAB_191221 - person - yolov3-tiny_prnYOLOv3-Tiny_PRN1. anchors2. yolov3-tiny_prn.data/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-master/data/yolov3-tiny_prn.dataclasses= 1train = /media/famu/DISK_DEEP/yongqiang/yolov3-ti

Darknet - 模型 (.weights) 重命名1. ******_final.weights / ******_last.weights 重命名******_bag_200016.weights******_bag_final.weights -> ******_bag_200026.weights******_bag_last.weights -> ******_bag_200036.weights2. rename_and_sort_weights_files.py#!

darknet_AlexeyAB_191221 - Makefile1. Makefile - source/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-masterGPU=0CUDNN=0CUDNN_HALF=0OPENCV=0AVX=0OPENMP=0LIBSO=0ZED_CAMERA=0# set GPU=1 and CUDNN=1 to speedup on GPU# set CUDNN_HALF=

linear_activate& linear_gradient1. darknet - activations.h -linear_activatestatic inline float linear_activate(float x){ return x;}2. darknet - activations.h -linear_gradientstatic inline float linear_gradient(float x){ return 1...

Darknet: Open Source Neural Networks in C - Tiny Darknethttps://pjreddie.com/darknet/Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation. You can find the source on

Darknet YOLO v3 Neural Network for object detection - 目标像素尺寸1. General rule - 一般规则Your training dataset should include such a set of relative sizes of objects that you want to detect:您的训练数据集应包含您要检测的一组相对大小的目标：train_network_width * train_obj_width / trai

Darknet - JPEGImages - labels 训练数据筛选可视化 (bag_selection)1. before2. Darknet - JPEGImages - labels 训练数据筛选可视化display_JPEGImages_labels_20191209_bag_selection.py#!/usr/bin/env python3# -*- coding: utf-8 -*-# Yongqiang Chengfrom __future__ import absol

OpenCV 4.2.0 环境下 AlexeyAB/darknet 编译和链接https://github.com/AlexeyAB/darknethttps://docs.opencv.org/4.2.0/index.html1. pjreddie/darknet - Makefile - OpenCV 2.x / 3.xGPU=1CUDNN=1CUDNN_HALF=0OPENCV=1AVX=0OPENMP=0LIBSO=0ZED_CAMERA=0# set GPU=1 and

Darknet: Open Source Neural Networks in C - Darknet Reference Modelhttps://pjreddie.com/darknet/Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation. You can find the

Validate Classification Model - Darknet validation log parser1. Darknet validation logyongqiang@famu-sys:~/darknet_work/darknet_181018/darknet$ bash ./backup_tiny_darknet_autoparts_001_valid_224.shlayer filters size input

Darknet - JPEGImages - labels 训练数据核验可视化 (bag)1. before2. Darknet - JPEGImages - labels 训练数据核验可视化display_JPEGImages_labels_20191209_bag.py#!/usr/bin/env python3# -*- coding: utf-8 -*-# Yongqiang Chengfrom __future__ import absolute_importfrom __fu

Python 2.7.15 / Python 3.6.5 - os.rename -深度学习数据集批量规范命名1. Python 2.7.15Python2 os.rename() 方法os.rename() 方法用于命名文件或目录，从 src 到 dst，如果 dst 是一个存在的目录，将抛出 OSError。src -- 要修改的目录名dst -- 修改后的目录名os.rename(src, dst)Rename the file or directory src to dst. I.

leaky_activate &leaky_gradient1. darknet - activations.h - leaky_activatestatic inline float leaky_activate(float x){ return (x > 0) ? x : .1 * x;}2.darknet - activations.h -leaky_gradientstatic inline float leaky_gradient(float x...

Darknet: Open Source Neural Networks in C - Classifying With darknet19.weights Models

pjreddie/darknetConvolutional Neural Networkshttps://github.com/pjreddie/darknethttps://pjreddie.com/darknet/Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.F

Visualizing and Understanding Convolutional Networks可视化和理解卷积网络 - 看懂卷积网络Matthew D Zeiler, Rob FergusDept. of Computer Science, Courant Institute, New York Universitydepartment [dɪ'pɑːtm(ə)nt]，dept....

ImageNet Classification with Deep Convolutional Neural NetworksAlex Krizhevsky, Ilya Sutskever, Geoffrey E. HintonUniversity of TorontoUniversity of Toronto，U of T, UToronto or Toronto：多伦多大学Abstr...

The Comprehensive Cars (CompCars) dataset - 车辆精细识别数据集 - surveillance-nature imagessurveillance-nature images(1) The surveillance data are released as “sv_data.*”.(2) sv_data.zip -> 解压到当前文件夹 -&a

BIT-Vehicle Dataset - 车辆车型识别数据集http://iitlab.bit.edu.cn/mcislab/vehicledb/Dataset infoThe BIT-Vehicle dataset contains 9,850 vehicle images. There are images with sizes of 1600*1200 and 1920*1080 ...

A guide to convolution arithmetic for deep learning - 深度学习卷积运算指南Vincent Dumoulin, Francesco VisinMILA, Université de MontréalAIRLab, Politecnico di MilanoThe Université de Montréal (UdeM) is a Fre...