Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/pprp/simplecvreproduction
Replication of simple CV Projects including attention, classification, detection, keypoint detection, etc.
https://github.com/pprp/simplecvreproduction
attention classification computer-vision cv demo face-detection landmark object-detection paper-reproduction pytorch
Last synced: 4 days ago
JSON representation
Replication of simple CV Projects including attention, classification, detection, keypoint detection, etc.
- Host: GitHub
- URL: https://github.com/pprp/simplecvreproduction
- Owner: pprp
- License: apache-2.0
- Created: 2019-12-11T12:25:06.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2023-01-08T14:13:48.000Z (almost 2 years ago)
- Last Synced: 2024-12-13T11:11:52.370Z (11 days ago)
- Topics: attention, classification, computer-vision, cv, demo, face-detection, landmark, object-detection, paper-reproduction, pytorch
- Language: Jupyter Notebook
- Homepage:
- Size: 253 MB
- Stars: 1,222
- Watchers: 25
- Forks: 311
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# SimpleCVReproduction
Recommended models are listed in this repository. In order to simplify learning for beginners, comments are provided with running models, code ready to read, well-documented code, and a series of simple code snippets. The aim of this project is to provide a simplified version of easy-to-understand model files. Suggestions for easily learnable libraries are welcome within the Issues section. Most of the content in this project is from Github and shall not be used for commercial purposes. In case of any infringement, please contact the author for removal.
将感兴趣/推荐的模型也放在这个库中，以供学习。由于好多库从头开始学习难度太大，在这里提供了笔者的部分注释，其中大部分都是跑过的模型、准备读的代码、已经读过的代码笔记、以及开发的simple系列简单代码、常用代码段等。
本项目致力于提供简化版本的，便于理解的模型文件。
如果有推荐的便于初学者学习的库，也欢迎在issue中提出和补充。
本项目大部分内容是来源于Github，不会用做商业用途，如有侵权，请联系笔者删除。
## 目录
- [即插即用模块&注意力模块](即插即用模块&注意力模块)
- [项目推荐](项目推荐)
- [致谢](致谢)
- [贡献](贡献)
## 即插即用模块&注意力模块
原项目已经迁移至新的地址：[Awesome-Attention-Mechanism-in-cv](https://github.com/pprp/awesome-attention-mechanism-in-cv)
主要内容包括:
- 计算机视觉领域中**注意力**模块。
- 计算机视觉中**即插即用**模块。[code](https://github.com/pprp/SimpleCVReproduction/tree/master/Plug-and-play%20module)
- **Vision Transformer**系列工作。
更多介绍：
- [我们是如何结合注意力机制改进YOLOv3进行目标检测?](https://zhuanlan.zhihu.com/p/231168560) [Code](https://github.com/GiantPandaCV/yolov3-point)
- [如何在YOLOv3中加入注意力模块or即插即用模块](https://blog.csdn.net/DD_PP_JJ/article/details/104109369)
- [神经网络加上注意力机制，精度反而下降，为什么会这样呢？](https://www.zhihu.com/question/478301531/answer/2280232845)
- [CNN、Transformer、MLP架构经验性分析](https://zhuanlan.zhihu.com/p/449280021)
- [CV中的注意力机制之ShuffleAttention](https://zhuanlan.zhihu.com/p/350912960)
- [CV中的注意力机制之并联版的CBAM-BAM模块](https://zhuanlan.zhihu.com/p/102033063)
- [CV中的注意力机制之SKNet-SENet的提升版](https://zhuanlan.zhihu.com/p/102034839)
- [CV中的注意力机制之简单而有效的CBAM模块](https://zhuanlan.zhihu.com/p/102035273)
- [CV中的注意力机制之SENet中的SE模块](https://zhuanlan.zhihu.com/p/102035721)
- [CV中的注意力机制之语义分割中的scSE模块](https://zhuanlan.zhihu.com/p/102036086)
- [CV中的注意力机制之Non-Local Network的理解与实现](https://zhuanlan.zhihu.com/p/102984842)
- [CV中的注意力机制之融合Non-Local和SENet的GCNet](https://zhuanlan.zhihu.com/p/102990363)
- [CV中的注意力机制之BiSeNet中的FFM模块与ARM模块](https://zhuanlan.zhihu.com/p/105925132)
## 项目推荐
| 项目 | 介绍 | 链接 |
| ------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
| CenterNet | 简化版本的CenterNet目标检测算法(第三方实现) | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/CenterNet) |
| SmallObjectAugmentation | 针对小目标进行数据增强库,在笔者数据集效果不理想 | [link](https://github.com/pprp/52RL) |
| DarkLabel | 专门用于[DarkLabel](https://zhuanlan.zhihu.com/p/141036498)软件转化的系列脚本 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/DarkLabel) |
| Latex/latex_algo | 用latex写的伪代码示例 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/Latex/latex_algo) |
| MLP | MLP-Mixer,ResMLP,RepMLP简单源码 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/MLP) |
| NAS | 感兴趣的神经网络结构搜索算法 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/NAS) |
| Plug-and-play Module | 即插即用模块 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/Plug-and-play%20module) |
| 52RL | 参加DataWhale深度强化学习课程代码 [code](https://github.com/pprp/52RL) | [link](https://github.com/pprp/52RL) |
| Vision Transformer | 最经典的ViT实现, 训练代码在[code](https://github.com/pprp/pytorch-cifar-model-zoo) | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/VisionTransformer) |
| captcha-CTC-loss | CTC loss+ LSTM | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/captcha-CTC-loss) |
| cifarTrick | 原先收集的部分Trick更多Trick在[Tricks](https://github.com/pprp/pytorch-cifar-model-zoo) | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/cifarTrick) |
| deep_sort | 官方实现的DeepSort算法 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/deep_sort) |
| deep_sort_yolov3_pytorch | 笔者自己实现和改进的DeepSort算法 | [link](https://github.com/pprp/deep_sort_yolov3_pytorch) |
| easy-receptive-fields | 感受野计算，分析，特征图可视化 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/easy-receptive-fields) |
| fine_grained_baseline | 细粒度识别baseline，Bilinear Pooling操作 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/fine_grained_baseline) |
| flask-yolo | flask配合yolo算法实现网页 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/flask-yolo) |
| kalman | 卡尔曼滤波实现与测试 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/kalman) |
| libfacedetection.train | 人脸检测训练代码 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/kalman) |
| opencv-mot | 使用Opencv实现多目标跟踪 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/opencv-mot) |
| pandoc-starter | Pandoc是Markdown转化器，很方便 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/pandoc-starter) |
| pytorch-commen-code | 常用的pytorch代码片段 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/pandoc-starter) |
| siamese-triplet | 孪生网络+Triplet Loss实现 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/simple-faster-rcnn-pytorch) |
| simple-triple-loss | 笔者自己实现的triplet loss | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/simple-triple-loss) |
| simple_keypoint | **[推荐]** 笔者极简代码实现关键点识别，提供根据heatmap进行识别的方法 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/simple_keypoint) |
| tikz_cnn | 使用latex绘制CNN图 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/tikz_cnn) |
| tsne | tsne可视化数据集 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/tsne) |
| tools | voc2coco脚本，yolo anchor聚类脚本 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/tools) |
| tiny_classifier | 超级简单的分类代码+focal loss使用 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/tiny_classifier) |
| yolov3-6 | 第六次release版本，属于老版本yolo实现 | [link](https://github.com/pprp/SimpleCVReproduction/tree/master/yolov3-6) |
## 致谢
@zhongqiu1245 补充的borderDet中的BAM模块,补充了FPT
@1187697147 补充的context-gating模块
@cmsfw-github 指出了simple_keypoint中的bug
@1187697147 建议更新了AFF和iAFF模块源码
## 贡献
欢迎在issue中提出补充推荐的项目。
欢迎关注“GiantPandaCV”公众号以及“神经网络架构搜索”公众号查看相关博客。