mobile-yolov5-pruning-distillation:mobilev2-yolov5s剪枝,蒸馏,支持ncnn,tensorRT部署。超轻但性能更好!-源码

上传者: 42165508 | 上传时间: 2021-03-30 20:19:06 | 文件大小: 5.91MB | 文件类型: ZIP
移动式yolov5修剪蒸馏 mobilev2-yolov5s的通道修剪和蒸馏。超轻但性能更好! TensorRT版本===> Android版本===> 背景 yolov5s在640x640分辨率下的计算量和参数量分别为8.39G和7.07M。在速度上仍然有提升空间,通过替换backbone(mobilenetv2),通道剪枝对模型进行压缩。 。本项目以工程化为基础,主要是模型端的优化。实现了常用的剪枝和蒸馏算法,并完成了一个简单的介绍和评估。将工程可用模型转换成对应的部署版本。 基准线 数据集采用Pascal VOC,训练集= train2007 + train2012 + val2007 + val2012,测试集= test2007,基线采用mobile-yolo(imagenet预训练),如果没有特别说明,第一个模块采用重点如果特定特殊说明则可以使用替换参数,batchsize

文件下载

资源详情

[{"title":"( 76 个子文件 5.91MB ) mobile-yolov5-pruning-distillation:mobilev2-yolov5s剪枝,蒸馏,支持ncnn,tensorRT部署。超轻但性能更好!-源码","children":[{"title":"mobile-yolov5-pruning-distillation-master","children":[{"title":"train.py <span style='color:#111;'> 34.81KB </span>","children":null,"spread":false},{"title":"utils","children":[{"title":"utils.py <span style='color:#111;'> 51.82KB </span>","children":null,"spread":false},{"title":"datasets.py <span style='color:#111;'> 34.19KB </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 0B </span>","children":null,"spread":false},{"title":"google_utils.py <span style='color:#111;'> 4.16KB </span>","children":null,"spread":false},{"title":"activations.py <span style='color:#111;'> 1.62KB </span>","children":null,"spread":false},{"title":"torch_utils.py <span style='color:#111;'> 8.34KB </span>","children":null,"spread":false}],"spread":true},{"title":"models","children":[{"title":"mobile-yolo5s.yaml <span style='color:#111;'> 2.11KB </span>","children":null,"spread":false},{"title":"yolov5m.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"yolov5l_voc.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"mobile-yolo5l_voc.yaml <span style='color:#111;'> 2.11KB </span>","children":null,"spread":false},{"title":"yolov5x.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"mobile-yolo5s_voc.yaml <span style='color:#111;'> 2.11KB </span>","children":null,"spread":false},{"title":"experimental.py <span style='color:#111;'> 3.30KB </span>","children":null,"spread":false},{"title":"yolov5s_voc.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 0B </span>","children":null,"spread":false},{"title":"yolo.py <span style='color:#111;'> 11.24KB </span>","children":null,"spread":false},{"title":"mobile-yolo3_voc.yaml <span style='color:#111;'> 2.56KB </span>","children":null,"spread":false},{"title":"yolov5x_voc.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"onnx_export.py <span style='color:#111;'> 1.67KB </span>","children":null,"spread":false},{"title":"yolov5s.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"yolov3-spp.yaml <span style='color:#111;'> 1.65KB </span>","children":null,"spread":false},{"title":"yolov5l.yaml <span style='color:#111;'> 1.34KB </span>","children":null,"spread":false},{"title":"common.py <span style='color:#111;'> 6.43KB </span>","children":null,"spread":false}],"spread":false},{"title":"test.py <span style='color:#111;'> 14.00KB </span>","children":null,"spread":false},{"title":"hubconf.py <span style='color:#111;'> 3.00KB </span>","children":null,"spread":false},{"title":"detect.py <span style='color:#111;'> 6.90KB </span>","children":null,"spread":false},{"title":"script","children":[{"title":"split_train_data.py <span style='color:#111;'> 909B </span>","children":null,"spread":false},{"title":"coco2darknet.py <span style='color:#111;'> 20.30KB </span>","children":null,"spread":false},{"title":"tfyolo2darknet.py <span style='color:#111;'> 1.96KB </span>","children":null,"spread":false},{"title":"labelme2darknet.py <span style='color:#111;'> 3.46KB </span>","children":null,"spread":false},{"title":"voc_label.py <span style='color:#111;'> 2.62KB </span>","children":null,"spread":false},{"title":"kmeans.py <span style='color:#111;'> 108B </span>","children":null,"spread":false},{"title":"cal_flop.py <span style='color:#111;'> 1017B </span>","children":null,"spread":false}],"spread":true},{"title":"Dockerfile <span style='color:#111;'> 1.74KB </span>","children":null,"spread":false},{"title":"pic","children":[{"title":"distillation_5.png <span style='color:#111;'> 123.39KB </span>","children":null,"spread":false},{"title":"distillation_2.png <span style='color:#111;'> 212.67KB </span>","children":null,"spread":false},{"title":"after_pruning_prob_05.jpg <span style='color:#111;'> 13.77KB </span>","children":null,"spread":false},{"title":"pruning_func.png <span style='color:#111;'> 70.60KB </span>","children":null,"spread":false},{"title":"distillation_1.jpg <span style='color:#111;'> 84.76KB </span>","children":null,"spread":false},{"title":"distillation_6.jpg <span style='color:#111;'> 59.40KB </span>","children":null,"spread":false},{"title":"demo4.jpg <span style='color:#111;'> 50.68KB </span>","children":null,"spread":false},{"title":"distillation_7.png <span style='color:#111;'> 66.20KB </span>","children":null,"spread":false},{"title":"after_pruning_thres_001.jpg <span style='color:#111;'> 13.19KB </span>","children":null,"spread":false},{"title":"pruning_func.jpg <span style='color:#111;'> 73.38KB </span>","children":null,"spread":false},{"title":"distillation_4.jpg <span style='color:#111;'> 35.13KB </span>","children":null,"spread":false},{"title":"distillation_3.png <span style='color:#111;'> 227.24KB </span>","children":null,"spread":false},{"title":"demo3.jpg <span style='color:#111;'> 50.03KB </span>","children":null,"spread":false},{"title":"demo1.jpg <span style='color:#111;'> 50.13KB </span>","children":null,"spread":false},{"title":"demo2.jpg <span style='color:#111;'> 51.04KB </span>","children":null,"spread":false}],"spread":false},{"title":"requirements.txt <span style='color:#111;'> 10B </span>","children":null,"spread":false},{"title":"weights","children":[{"title":"download_weights.sh <span style='color:#111;'> 244B </span>","children":null,"spread":false}],"spread":true},{"title":"LICENSE <span style='color:#111;'> 1.04KB </span>","children":null,"spread":false},{"title":"inference","children":[{"title":"output","children":[{"title":"bus.jpg <span style='color:#111;'> 474.01KB </span>","children":null,"spread":false},{"title":"zidane.jpg <span style='color:#111;'> 223.59KB </span>","children":null,"spread":false},{"title":"coco_1.jpg <span style='color:#111;'> 101.38KB </span>","children":null,"spread":false}],"spread":true},{"title":"images","children":[{"title":"bus.jpg <span style='color:#111;'> 476.01KB </span>","children":null,"spread":false},{"title":"zidane.jpg <span style='color:#111;'> 164.99KB </span>","children":null,"spread":false}],"spread":true}],"spread":true},{"title":"README.md <span style='color:#111;'> 16.34KB </span>","children":null,"spread":false},{"title":"outputs","children":[{"title":"voc","children":[{"title":"train_batch0.jpg <span style='color:#111;'> 145.97KB </span>","children":null,"spread":false},{"title":"train_batch1.jpg <span style='color:#111;'> 259.06KB </span>","children":null,"spread":false},{"title":"train_batch2.jpg <span style='color:#111;'> 113.08KB </span>","children":null,"spread":false},{"title":"label.png <span style='color:#111;'> 529.23KB </span>","children":null,"spread":false},{"title":"results.txt <span style='color:#111;'> 7.37KB </span>","children":null,"spread":false}],"spread":false}],"spread":true},{"title":"tutorial.ipynb <span style='color:#111;'> 3.11MB </span>","children":null,"spread":false},{"title":"data","children":[{"title":"coco.yaml <span style='color:#111;'> 1.63KB </span>","children":null,"spread":false},{"title":"voc.yaml <span style='color:#111;'> 343B </span>","children":null,"spread":false},{"title":"coco128.yaml <span style='color:#111;'> 1.47KB </span>","children":null,"spread":false},{"title":"get_coco2017.sh <span style='color:#111;'> 975B </span>","children":null,"spread":false}],"spread":false},{"title":"torch_pruning","children":[{"title":"utils.py <span style='color:#111;'> 1.27KB </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 66B </span>","children":null,"spread":false},{"title":"prune","children":[{"title":"unstructured.py <span style='color:#111;'> 1.68KB </span>","children":null,"spread":false},{"title":"structured.py <span style='color:#111;'> 6.12KB </span>","children":null,"spread":false},{"title":"__init__.py <span style='color:#111;'> 53B </span>","children":null,"spread":false}],"spread":false},{"title":"dependency.py <span style='color:#111;'> 18.49KB </span>","children":null,"spread":false}],"spread":false},{"title":"pruning.py <span style='color:#111;'> 5.51KB </span>","children":null,"spread":false}],"spread":false}],"spread":true}]

评论信息

  • weiqiy :
    用户下载后在一定时间内未进行评价,系统默认好评。
    2021-08-21
  • weixin_43269994 :
    用户下载后在一定时间内未进行评价,系统默认好评。
    2021-08-12
  • zhuangwu116 :
    用户下载后在一定时间内未进行评价,系统默认好评。
    2021-08-10

免责申明

【只为小站】的资源来自网友分享,仅供学习研究,请务必在下载后24小时内给予删除,不得用于其他任何用途,否则后果自负。基于互联网的特殊性,【只为小站】 无法对用户传输的作品、信息、内容的权属或合法性、合规性、真实性、科学性、完整权、有效性等进行实质审查;无论 【只为小站】 经营者是否已进行审查,用户均应自行承担因其传输的作品、信息、内容而可能或已经产生的侵权或权属纠纷等法律责任。
本站所有资源不代表本站的观点或立场,基于网友分享,根据中国法律《信息网络传播权保护条例》第二十二条之规定,若资源存在侵权或相关问题请联系本站客服人员,zhiweidada#qq.com,请把#换成@,本站将给予最大的支持与配合,做到及时反馈和处理。关于更多版权及免责申明参见 版权及免责申明