yolov5_violence_detect

基于yolov5s进行的暴力行为检测

https://github.com/tcccth/yolov5_violence_detect

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

✓
CITATION.cff file
Found CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
✓
.zenodo.json file
Found .zenodo.json file
○
DOI references
○
Academic publication links
○
Academic email domains
○
Institutional organization owner
○
JOSS paper metadata
○
Scientific vocabulary similarity
Low similarity (1.3%) to scientific vocabulary

Last synced: 10 months ago · JSON representation ·

Repository

基于yolov5s进行的暴力行为检测

Basic Info

Host: GitHub
Owner: tcccth
Language: Python
Default Branch: master
Size: 27.2 MB

Statistics

Stars: 9
Watchers: 1
Forks: 1
Open Issues: 1
Releases: 0

Created almost 2 years ago · Last pushed almost 2 years ago

Metadata Files

Readme Citation

README.md

一. 模型修改

1. 使用模型

本次训练使用了 yolov5s 模型，所以需要对部分配置文件进行修改：

需要对位于文件夹内 ./models/yolov5s.yaml 进行修改，训练修改了其的复制文件：

根据自身数据集标签数量对 nc 进行修改。

同时，对 ./data/train.yaml 进行修改：

最后修改训练文件：

并且调整其中参数。

二. 训练数据

1. 性能指标分析

(1) 模型数据

训练好的模型保存在 ./runs/train/exp2/weights 中

其分析数据存放于同级文件夹下

(2). 数据分析

lables 反映了训练集的数据量，以及其图片和标签的部分参数

results 反映了训练集和数据集的结果，包含了：训练次数，GPU消耗，边界框损失，目标检测损失，total，targets, 图片大小，P，R，mAP@.5, mAP@.5:.95, 验证集val Box, 验证集val obj, 验证集val cls

上图为 train.csv

同时，也可以通过损失函数评估模型性能。

定位损失box_loss：预测框与标定框之间的误差（GIoU）
置信度损失obj_loss：计算网络的置信度
分类损失cls_loss：计算锚框与对应的标定分类是否正确

上图为train.png

特别注意到：在最后数轮次的训练中，验证集的损失有所上升，可能需要更换置信度损失

其他曲线趋于稳定，cls_loss则始终为0，可能是class设置不多，模型训练未将其丢失

(3). 模型构成

于pycharm终端，输入 tensorboard --logdir runs，将结果可视化

通过浏览器查看可视化训练结果

于其中查看本次模型的构成：

可以看到，模型通过多层卷积，采样，实现训练。

上图yolov5s的代码部分

训练结果

模型将会检测视频或图片中的暴力行为

注意到：模型在面对人体剧烈姿态时，可能会出现假正例，以及部分假负例。可能是模型对人体姿态和二者距离之间产生错误联系。需要对数据集标签重新精确标定，并且加大数据集训练

Owner

Login: tcccth
Kind: user

Repositories: 1
Profile: https://github.com/tcccth

Citation (CITATION.cff)

cff-version: 1.2.0
preferred-citation:
  type: software
  message: If you use YOLOv5, please cite it as below.
  authors:
  - family-names: Jocher
    given-names: Glenn
    orcid: "https://orcid.org/0000-0001-5950-6979"
  title: "YOLOv5 by Ultralytics"
  version: 7.0
  doi: 10.5281/zenodo.3908559
  date-released: 2020-5-29
  license: GPL-3.0
  url: "https://github.com/ultralytics/yolov5"

GitHub Events

Total

Issues event: 1
Watch event: 10

Last Year

Issues event: 1
Watch event: 10

Issues and Pull Requests

Last synced: 10 months ago

All Time

Total issues: 1
Total pull requests: 0
Average time to close issues: N/A
Average time to close pull requests: N/A
Total issue authors: 1
Total pull request authors: 0
Average comments per issue: 0.0
Average comments per pull request: 0
Merged pull requests: 0
Bot issues: 0
Bot pull requests: 0

Past Year

Issues: 1
Pull requests: 0
Average time to close issues: N/A
Average time to close pull requests: N/A
Issue authors: 1
Pull request authors: 0
Average comments per issue: 0.0
Average comments per pull request: 0
Merged pull requests: 0
Bot issues: 0
Bot pull requests: 0

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Open Source Science