-
YOLOv5
https://github.com/ultralytics/yolov5 https://github.com/ultralytics/yolov5 ; -
YOLOv5-Multibackbone-Compression https://github.com/Gumpest/YOLOv5-Multibackbone-Compression ;
2021.10.30 复现TPH-YOLOv5
2021.10.31 完成替换backbone为Ghostnet
2021.11.02 完成替换backbone为Shufflenetv2
2021.11.05 完成替换backbone为Mobilenetv3Small
2021.11.10 完成EagleEye对YOLOv5系列剪枝支持
2021.11.14 完成MQBench对YOLOv5系列量化支持
2021.11.16 完成替换backbone为EfficientNetLite-0
2021.11.26 完成替换backbone为PP-LCNet-1x
2021.12.12 完成SwinTrans-YOLOv5(C3STR)
2021.12.15 完成Slimming对YOLOv5系列剪枝支持
3.
百度安全验证https://baijiahao.baidu.com/s?id=1713695277474830019&wfr=spider&for=pc ;
Perform a series of ablation experiments on yolov5 to make it lighter (smaller Flops, lower memory, and fewer parameters) and faster (add shuffle channel, yolov5 head for channel reduce. It can infer at least 10+ FPS On the Raspberry Pi 4B when input the frame with 320×320) and is easier to deploy (removing the Focus layer and four slice operations, reducing the model quantization accuracy to an acceptable range).
在 yolov5 上进行一系列消融实验,使其更轻(更小的 Flops、更低的内存和更少的参数)和更快(添加 shuffle channel,yolov5 head for channel reduce。它可以在 Raspberry Pi 4B 上推断至少 10+ FPS 时 输入320×320的帧)更容易部署(去掉Focus层和四片操作,将模型量化精度降低到可以接受的范围内)。
https://github.com/deepcam-cn/yolov5-face https://github.com/deepcam-cn/yolov5-face ;8.旋转目标检测
https://github.com/BossZard/rotation-yolov5 https://github.com/BossZard/rotation-yolov5 ;
https://github.com/hukaixuan19970627/yolov5_obb https://github.com/hukaixuan19970627/yolov5_obb ;9. 剪枝蒸馏
https://github.com/xhwNobody/yolov5_prune_sfp https://github.com/xhwNobody/yolov5_prune_sfp ; https://github.com/mvpzhangqiu/yolov5prune https://github.com/mvpzhangqiu/yolov5prune ;
https://github.com/ZJU-lishuang/yolov5_prune https://github.com/ZJU-lishuang/yolov5_prune ; GitHub – midasklr/yolov5prune Contribute to midasklr/yolov5prune development by creating an account on GitHub.https://github.com/midasklr/yolov5prune ; GitHub – jiachengjiacheng/Pruned-YOLO: Using model pruning method to obtain compact models Pruned-YOLOv5 based on YOLOv5.Using model pruning method to obtain compact models Pruned-YOLOv5 based on YOLOv5. – GitHub – jiachengjiacheng/Pruned-YOLO: Using model pruning method to obtain compact models Pruned-YOLOv5 based on YOLOv5.https://github.com/jiachengjiacheng/Pruned-YOLO ;
https://github.com/Syencil/mobile-yolov5-pruning-distillation https://github.com/Syencil/mobile-yolov5-pruning-distillation ; yolov5 (V6) 剪枝的干净代码版本:
- The original Yolo V5 was an amazing project. When I want to make some changes to the network, it’s not so easy, such as adding branches and trying other backbones. Maybe there are people like me, so I split the yolov5 model to {backbone, neck, head} to facilitate the operation of various modules and support more backbones.Basically, I only changed the model, and I didn’t change the architecture, training and testing of yolov5. Therefore, if the original code is updated, it is also very convenient to update this code. if this repo can help you, please give me a star.
最初的 Yolo V5 是一个了不起的项目。 当我想对网络进行一些更改时,并不是那么容易,例如添加分支和尝试其他主干。 可能有我这样的人,所以我把yolov5模型拆分成{backbone,neck,head},方便各个模块的操作,支持更多的主干。基本上我只改了模型,没改架构, yolov5的训练和测试。 因此,如果更新原始代码,更新这段代码也很方便。 如果这个repo可以帮助你,请给我一个star。
GitHub – cv516Buaa/tph-yolov5 https://github.com/cv516Buaa/tph-yolov5 ;12. yolov5注释版
GitHub – SCAU-HuKai/yolov5-5.x-annotations: 一个基于yolov5-5.0的中文注释版本!https://github.com/SCAU-HuKai/yolov5-5.x-annotations ;
https://github.com/Laughing-q/yolov5_annotations https://github.com/Laughing-q/yolov5_annotations ;13. 车牌检测
https://github.com/zeusees/License-Plate-Detector https://github.com/zeusees/License-Plate-Detector ;14. YOLODet-PyTorch是端到端基于pytorch框架复现yolo最新算法的目标检测开发套件,旨在帮助开发者更快更好地完成检测模型的训练、精度速度优化到部署全流程。YOLODet-PyTorch以模块化的设计实现了多种主流YOLO目标检测算法,并且提供了丰富的数据增强、网络组件、损失函数等模块。
https://github.com/wuzhihao7788/yolodet-pytorch https://github.com/wuzhihao7788/yolodet-pytorch ;15. 蒸馏
GitHub – Sharpiless/yolov5-distillation-5.0: yolov5 5.0 version distillation || yolov5 5.0版本知识蒸馏,yolov5l >> yolov5s yolov5 5.0 version distillation || yolov5 5.0版本知识蒸馏,yolov5l >> yolov5s – GitHub – Sharpiless/yolov5-distillation-5.0: yolov5 5.0 version distillation || yolov5 5.0版本知识蒸馏,yolov5l >> yolov5s https://github.com/Sharpiless/yolov5-distillation-5.0 ; GitHub – Sharpiless/yolov5-knowledge-distillation: yolov5目标检测模型的知识蒸馏(基于响应的蒸馏)https://github.com/Sharpiless/yolov5-knowledge-distillation ; GitHub – SsisyphusTao/Object-Detection-Knowledge-Distillation: An Object Detection Knowledge Distillation framework powered by pytorch, now having SSD and yolov5.An Object Detection Knowledge Distillation framework powered by pytorch, now having SSD and yolov5. – GitHub – SsisyphusTao/Object-Detection-Knowledge-Distillation: An Object Detection Knowledge Distillation framework powered by pytorch, now having SSD and yolov5.https://github.com/SsisyphusTao/Object-Detection-Knowledge-Distillation ;
GitHub – Sharpiless/Yolov5-distillation-train-inference: Yolov5 distillation training | Yolov5知识蒸馏训练,支持训练自己的数据https://github.com/Sharpiless/Yolov5-distillation-train-inference ;16. yolov5 with bi-fpn
accomplished
- 2021.12.15
- change backbone to Ghostnet
- Finish EagleEye pruning YOLOv5 series
- 2021.12.27
- change backbone to shufflenetv2
- change backbone to efficientnetv2
https://github.com/Ranking666/Yolov5-Processing https://github.com/Ranking666/Yolov5-Processing ;22. transformer-yolov5
https://github.com/pat-langevin/yolov5_transformer https://github.com/pat-langevin/yolov5_transformer ;23.可以训练yolov5(v6.0)、yolox、小型网络,添加注意力机制
https://github.com/xialuxi/yolov5_all https://github.com/xialuxi/yolov5_all ;24.
- 小麦检测,添加mix-up数据增强
- 替换 YOLOv5 tag=4.0 主干:MobileNetV2@1.0 和 MobileNetV2@0.25
https://github.com/RedHandLM/MobileNet-YOLOv5 https://github.com/RedHandLM/MobileNet-YOLOv5 ;27.
- 在yolox正添加训练yolov5的数据格式,以及测试代码。
- 目前是在单gpu模式下修改调试的代码,多gpu下还需要。
- 具体修改:
- 在yolox/data/datasets/下添加yolo.py和yolo_classes.py
- 在yolox/data/datasets/ init .py中添加
- 从 .yolo 导入 YOLODataset
- 从 .yolo_classes 导入 YOLO_CLASSES
- 在yolox/evaluators/下添加yolo_evaluator.py
- 在yolox/evaluators/ init .py中添加:
- 从 .yolo_evaluator 导入 YOLOEvaluator
- 训练的配置文件参考yolox_yolo_s.py
- 修改代码,多gpu下分别计算各自卡上的map,未做多gpu同步计算。
- 增加多gpu计算map的代码
GitHub – xialuxi/yolox-yolov5 https://github.com/xialuxi/yolox-yolov5 ;28.mobilenetv3,shufflenetv2
This repo code uses Softmax and CrossEntropyLoss instead of BCEWithLogitsLoss.The improved source code is fully compatible with the original version of YOLOv5:v5. At the same time, backbone supports mobilenetv3,shufflenetv2, the original backbone supports all, and so on.
2020.12.1:
- 修改 RetinaNet/FCOS 损失计算方法。训练时间减少 40%,提高模型性能。
2021.5.18:
- 所有分类/检测/分割模型在 tools/ 中都有一个公共的 train.py 和 test.py 文件。
- 对于训练和测试,分别在工作目录中生成 train.info.log 和 test.info.log 文件。
- 在 simpleAICV/classification/backbones/repvgg.py 中构建 repvgg 网络。GitHub – zgcr/simpleAICV-pytorch-ImageNet-COCO-training: Training examples and results for ImageNet(ILSVRC2012)/COCO2017/VOC2007+VOC2012 datasets.Include ResNet/DarkNet/RegNet/RetinaNet/FCOS/CenterNet/YOLO series.https://github.com/zgcr/simpleAICV-pytorch-ImageNet-COCO-training ;31.基于UltralyTICS/YOLOV5对repository进行了重构和标注,并增加了Grab Cut自动标注、指向中心点等功能。 该存储基于Ultralytics/yolov5进行补充与注释,并在此自动标记基础与填充GrabCut,以及绘制库中心点进行连线。https://github.com/isLinXu/YOLOv5_Efficient https://github.com/isLinXu/YOLOv5_Efficient ; 32. 目标检测yolov5 v6.0版,包含了目标检测数据标注,数据集增强,自定义训练集全流程。
- https://github.com/ami66/yolov5_v6.0_object_detection https://github.com/ami66/yolov5_v6.0_object_detection ;33. 使用伪标签
- GitHub – kimy-de/YOLOv5-PseudoLabels: YOLOv5 with PseudoLabels https://github.com/kimy-de/YOLOv5-PseudoLabels ;34.注意力
- https://github.com/csfenghan/Improved-Yolov5 https://github.com/csfenghan/Improved-Yolov5 ;
35.
这项工作通过我们提出的平衡焦点损失和损失等级挖掘(LRM)方法的组合修改了原始的YOLOv5工作以进行困难示例挖掘。原始loss.py文件已针对 LRM 实施进行了修改。
在v5.0 版本的 YOLOv5s 上测试并在v6.0 版本的 YOLOv5l 上验证。
使用 LRM_ignore 标志进行 LRM 激活,使用 fl_gamma 和 obj 标志进行平衡焦点损失。
相关论文:http ://arxiv.org/abs/2202.13080
https://github.com/aybora/yolov5Loss https://github.com/aybora/yolov5Loss ;36. 实例分割
https://github.com/jkd2021/YOLOv5-Model-with-Lane-Detection https://github.com/jkd2021/YOLOv5-Model-with-Lane-Detection ;38. 在”models”文件夹中添加dcn和4个头模型配置文件以处理不同大小的对象
39.基于yolov5的改进库
https://github.com/ppogg/YOLOv5-Lite https://github.com/ppogg/YOLOv5-Lite ;41.PP-LCnet
https://github.com/OutBreak-hui/Yolov5-PP-LCNet https://github.com/OutBreak-hui/Yolov5-PP-LCNet ;
Original: https://blog.csdn.net/dream_passionately/article/details/123412588
Author: dream_passionately
Title: yolo项目汇总
原创文章受到原创版权保护。转载请注明出处:https://www.johngo689.com/533447/
转载文章受原作者版权保护。转载请注明原作者出处!