yolov5目标检测模型的知识蒸馏（基于响应的蒸馏）

Last update: Dec 04, 2022

Overview

代码地址：

https://github.com/Sharpiless/yolov5-knowledge-distillation

教师模型：

python train.py --weights weights/yolov5m.pt \
        --cfg models/yolov5m.yaml --data data/voc.yaml --epochs 50 \
        --batch-size 8 --device 0 --hyp data/hyp.scratch.yaml

蒸馏训练：

python train.py --weights weights/yolov5s.pt \
        --cfg models/yolov5s.yaml --data data/voc.yaml --epochs 50 \
        --batch-size 8 --device 0 --hyp data/hyp.scratch.yaml \
        --t_weights yolov5m.pt --distill

训练参数:

--weights：预训练模型

--t_weights：教师模型权重

--distill：使用知识蒸馏进行训练

--dist_loss：l2或者kl

--temperature：使用知识蒸馏时的温度

使用《Object detection at 200 Frames Per Second》中的损失

这篇文章分别对这几个损失函数做出改进，具体思路为只有当teacher network的objectness value高时，才学习bounding box坐标和class probabilities。

实验结果：

这里假设VOC2012中新增加的数据为无标签数据（2k张）。

教师模型	训练方法	蒸馏损失	P	R	mAP50
无	正常训练	不使用	0.7756	0.7115	0.7609
Yolov5l	output based	l2	0.7585	0.7198	0.7644
Yolov5l	output based	KL	0.7417	0.7207	0.7536
Yolov5m	output based	l2	0.7682	0.7436	0.7976
Yolov5m	output based	KL	0.7731	0.7313	0.7931

参数和细节正在完善，支持KL散度、L2 logits损失和Sigmoid蒸馏损失等

1. 正常训练：

2. L2蒸馏损失：

我的公众号：

关于作者

B站：https://space.bilibili.com/470550823

CSDN：https://blog.csdn.net/weixin_44936889

AI Studio：https://aistudio.baidu.com/aistudio/personalcenter/thirdview/67156

Github：https://github.com/Sharpiless

yolov5目标检测模型的知识蒸馏（基于响应的蒸馏）

Related tags

Overview

代码地址：

教师模型：

蒸馏训练：

训练参数:

实验结果：

1. 正常训练：

2. L2蒸馏损失：

我的公众号：

关于作者

Owner

Investigating automatic navigation towards standard US views integrating MARL with the virtual US environment developed in CT2US simulation

A model which classifies reviews as positive or negative.

Pytorch implementation for "Implicit Semantic Response Alignment for Partial Domain Adaptation"

A `Neural = Symbolic` framework for sound and complete weighted real-value logic

buildseg is a building extraction plugin of QGIS based on PaddlePaddle.

A pytorch implementation of MBNET: MOS PREDICTION FOR SYNTHESIZED SPEECH WITH MEAN-BIAS NETWORK

Adversarial vulnerability of powerful near out-of-distribution detection

Variational autoencoder for anime face reconstruction

Semiconductor Machine learning project

Code for "CloudAAE: Learning 6D Object Pose Regression with On-line Data Synthesis on Point Clouds" @ICRA2021

Extreme Dynamic Classifier Chains - XGBoost for Multi-label Classification

This is a deep learning-based method to segment deep brain structures and a brain mask from T1 weighted MRI.

Code Repo for the ACL21 paper "Common Sense Beyond English: Evaluating and Improving Multilingual LMs for Commonsense Reasoning"

Implementation of Barlow Twins paper

Incorporating Transformer and LSTM to Kalman Filter with EM algorithm

COIN the currently largest dataset for comprehensive instruction video analysis.

Plover-tapey-tape: an alternative to Plover’s built-in paper tape

Official implementation for "Symbolic Learning to Optimize: Towards Interpretability and Scalability"

[CVPR'21] Projecting Your View Attentively: Monocular Road Scene Layout Estimation via Cross-view Transformation

An experimental technique for efficiently exploring neural architectures.