# Differentiable NAS for OSNet-AIN
## Introduction
This repository contains the neural architecture search (NAS) code (based on [Torchreid](https://arxiv.org/abs/1910.10093)) for [OSNet-AIN](https://arxiv.org/abs/1910.06827), an extension of [OSNet](https://arxiv.org/abs/1905.00953) that achieves strong performance on cross-domain person re-identification (re-ID) benchmarks (*without using any target data*). OSNet-AIN builds on the idea of using [instance normalisation](https://arxiv.org/abs/1607.08022) (IN) layers to eliminate instance-specific contrast in images for domain-generalisable representation learning. This is inspired by the [neural style transfer](https://arxiv.org/abs/1703.06868) works that use IN to remove image styles. Though IN naturally suits the cross-domain person re-ID task, it still remains unclear that where to insert IN to a re-ID CNN can maximise the performance gain. To avoid exhaustively evaluating all possible designs, OSNet-AIN learns to search for the optimal OSNet+IN design from data using a differentiable NAS algorithm. For technical details, please refer to our paper at https://arxiv.org/abs/1910.06827.
<div align="center">
<img src="https://drive.google.com/uc?export=view&id=1yvVIi2Ml7WBe85Uhaa54qyG4g8z-MGEB" width="500px" />
</div>
## Training
Assume the reid data is stored at `$DATA`. Run
```
python main.py --config-file nas.yaml --root $DATA
```
The structure of the found architecture will be shown at the end of training.
The default config was designed for 8 Tesla V100 32GB GPUs. You can modify the batch size based on your device memory.
**Note** that the test result obtained at the end of architecture search is not meaningful (due to the stochastic sampling layers). Therefore, do not rely on the result to judge the model performance. Instead, you should construct the found architecture in `osnet_child.py` and re-train and evaluate the model on the reid datasets.
## Citation
If you find this code useful to your research, please consider citing the following papers.
```
@article{zhou2021osnet,
title={Learning Generalisable Omni-Scale Representations for Person Re-Identification},
author={Zhou, Kaiyang and Yang, Yongxin and Cavallaro, Andrea and Xiang, Tao},
journal={TPAMI},
year={2021}
}
@inproceedings{zhou2019osnet,
title={Omni-Scale Feature Learning for Person Re-Identification},
author={Zhou, Kaiyang and Yang, Yongxin and Cavallaro, Andrea and Xiang, Tao},
booktitle={ICCV},
year={2019}
}
```
没有合适的资源?快使用搜索试试~ 我知道了~
基于YOLOv7和deepsoort实现目标检测(源码+模型).rar
共500个文件
py:178个
jpg:144个
yaml:65个
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
5星 · 超过95%的资源 8 下载量 38 浏览量
2023-04-15
20:17:10
上传
评论 4
收藏 35.93MB RAR 举报
温馨提示
1、资源内容:基于YOLOv7和deepsoort实现目标检测(源码+模型).rar 2、适用人群:计算机,电子信息工程、数学等专业的大学生课程设计、期末大作业或毕业设计,作为“参考资料”使用。 3、更多仿真源码和数据集下载列表(自行寻找自己需要的):https://blog.csdn.net/m0_62143653?type=download 4、免责声明:本资源作为“参考资料”而不是“定制需求”不一定能够满足所有人的需求,需要有一定的基础能够看懂代码,能够自行调试,能够自行添加功能修改代码。由于作者大厂工作较忙,不提供答疑服务,如不存在资源缺失问题概不负责,谢谢理解。
资源推荐
资源详情
资源评论
收起资源包目录
基于YOLOv7和deepsoort实现目标检测(源码+模型).rar (500个子文件)
events.out.tfevents.1665048173.cjc.22166.0 12KB
events.out.tfevents.1664958411.cjc.7631.0 8KB
events.out.tfevents.1664968115.cjc.8371.0 6KB
events.out.tfevents.1665067478.cjc.3681.0 6KB
events.out.tfevents.1664898891.cjc.22151.0 5KB
events.out.tfevents.1665062176.cjc.26522.0 5KB
events.out.tfevents.1665067452.cjc.3496.0 5KB
events.out.tfevents.1665048140.cjc.21982.0 5KB
events.out.tfevents.1665062290.cjc.30010.0 5KB
events.out.tfevents.1665062152.cjc.26356.0 5KB
events.out.tfevents.1665062192.cjc.26686.0 5KB
events.out.tfevents.1665054975.cjc.25184.0 4KB
events.out.tfevents.1665038646.cjc.13514.0 2KB
events.out.tfevents.1664985250.cjc.15046.0 1KB
events.out.tfevents.1664922774.cjc.84158.0 693B
events.out.tfevents.1664930254.cjc.88317.0 693B
events.out.tfevents.1664977715.cjc.11443.0 125B
events.out.tfevents.1664955836.cjc.7361.0 125B
events.out.tfevents.1664980261.cjc.11762.0 125B
events.out.tfevents.1664975822.cjc.10927.0 125B
events.out.tfevents.1664983031.cjc.13813.0 125B
events.out.tfevents.1664982760.cjc.13419.0 125B
events.out.tfevents.1664985189.cjc.14916.0 125B
events.out.tfevents.1664973186.cjc.9900.0 125B
events.out.tfevents.1664982781.cjc.13514.0 125B
events.out.tfevents.1665047656.cjc.21420.0 40B
events.out.tfevents.1665048055.cjc.21749.0 40B
events.out.tfevents.1665038295.cjc.13373.0 40B
events.out.tfevents.1665048083.cjc.21848.0 40B
events.out.tfevents.1665038150.cjc.13313.0 40B
events.out.tfevents.1665047582.cjc.21300.0 40B
events.out.tfevents.1665047719.cjc.21533.0 40B
events.out.tfevents.1665038045.cjc.13262.0 40B
events.out.tfevents.1665038361.cjc.13428.0 40B
events.out.tfevents.1665037626.cjc.12205.0 40B
events.out.tfevents.1664982938.cjc.13717.0 40B
events.out.tfevents.1664983006.cjc.13766.0 40B
events.out.tfevents.1665048128.cjc.21946.0 40B
events.out.tfevents.1664982829.cjc.13641.0 40B
events.out.tfevents.1665058696.cjc.25933.0 40B
events.out.tfevents.1664922323.cjc.81552.0 40B
.isort.cfg 307B
gnn_propagate_kernel.cu 1KB
build_adjacency_matrix_kernel.cu 1KB
.flake8 496B
.gitkeep 0B
heisi.jpg 254KB
train_batch4.jpg 248KB
train_batch6.jpg 246KB
train_batch8.jpg 246KB
train_batch0.jpg 245KB
train_batch2.jpg 239KB
train_batch9.jpg 235KB
train_batch1.jpg 229KB
train_batch5.jpg 224KB
train_batch3.jpg 218KB
train_batch7.jpg 201KB
train_batch8.jpg 190KB
ranking_results.jpg 186KB
train_batch3.jpg 181KB
train_batch1.jpg 176KB
train_batch8.jpg 168KB
train_batch6.jpg 164KB
train_batch6.jpg 164KB
train_batch6.jpg 164KB
train_batch6.jpg 164KB
train_batch6.jpg 164KB
train_batch6.jpg 164KB
train_batch5.jpg 159KB
train_batch2.jpg 158KB
train_batch9.jpg 155KB
train_batch3.jpg 152KB
train_batch3.jpg 152KB
train_batch1.jpg 149KB
train_batch4.jpg 147KB
train_batch4.jpg 147KB
train_batch0.jpg 146KB
train_batch7.jpg 143KB
train_batch4.jpg 143KB
train_batch4.jpg 142KB
train_batch4.jpg 141KB
train_batch4.jpg 141KB
train_batch4.jpg 141KB
train_batch4.jpg 141KB
train_batch4.jpg 141KB
train_batch5.jpg 140KB
train_batch5.jpg 140KB
train_batch7.jpg 139KB
girl.jpg 139KB
train_batch6.jpg 136KB
train_batch8.jpg 132KB
train_batch8.jpg 132KB
train_batch5.jpg 130KB
train_batch7.jpg 130KB
train_batch7.jpg 130KB
train_batch1.jpg 130KB
train_batch8.jpg 129KB
train_batch8.jpg 129KB
train_batch8.jpg 129KB
train_batch8.jpg 129KB
共 500 条
- 1
- 2
- 3
- 4
- 5
Matlab仿真实验室
- 粉丝: 2w+
- 资源: 2180
下载权益
C知道特权
VIP文章
课程特权
开通VIP
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
- 1
- 2
前往页