# KLD
> [Learning High-Precision Bounding Box for Rotated Object Detection via Kullback-Leibler Divergence](https://arxiv.org/pdf/2106.01883.pdf)
<!-- [ALGORITHM] -->
## Abstract
<div align=center>
<img src="https://raw.githubusercontent.com/zytx121/image-host/main/imgs/kld.png" width="800"/>
</div>
Existing rotated object detectors are mostly inherited from the horizontal detection paradigm, as the latter has evolved into a well-developed area. However, these detectors are difficult to perform prominently in high-precision detection due to the limitation of current regression loss design, especially for objects with large aspect ratios. Taking the perspective that horizontal detection is a special case for rotated object detection, in this paper, we are motivated to change the design of rotation regression loss from induction paradigm to deduction methodology, in terms of the relation between rotation and horizontal detection. We show that one essential challenge is how to modulate the coupled parameters in the rotation regression loss, as such the estimated parameters can influence to each other during the dynamic joint optimization, in an adaptive and synergetic way. Specifically, we first convert the rotated bounding box into a 2-D Gaussian distribution, and then calculate the Kullback-Leibler Divergence (KLD) between the Gaussian distributions as the regression loss. By analyzing the gradient of each parameter, we show that KLD (and its derivatives) can dynamically adjust the parameter gradients according to the characteristics of the object. For instance, it will adjust the importance (gradient weight) of the angle parameter according to the aspect ratio. This mechanism can be vital for high-precision detection as a slight angle error would cause a serious accuracy drop for large aspect ratios objects. More importantly, we have proved that KLD is scale invariant. We further show that the KLD loss can be degenerated into the popular $l\_{n}$-norm loss for horizontal detection. Experimental results on seven datasets using different detectors show its consistent superiority
## Results and models
DOTA1.0
| Backbone | mAP | Angle | lr schd | Mem (GB) | Inf Time (fps) | Aug | Batch Size | Configs | Download |
| :----------------------: | :---: | :---: | :-----: | :------: | :------------: | :-: | :--------: | :-----------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| ResNet50 (1024,1024,200) | 64.55 | oc | 1x | 3.38 | 15.7 | - | 2 | [rotated_retinanet_hbb_r50_fpn_1x_dota_oc](../rotated_retinanet/rotated_retinanet_hbb_r50_fpn_1x_dota_oc.py) | [model](https://download.openmmlab.com/mmrotate/v0.1.0/rotated_retinanet/rotated_retinanet_hbb_r50_fpn_1x_dota_oc/rotated_retinanet_hbb_r50_fpn_1x_dota_oc-e8a7c7df.pth) \| [log](https://download.openmmlab.com/mmrotate/v0.1.0/rotated_retinanet/rotated_retinanet_hbb_r50_fpn_1x_dota_oc/rotated_retinanet_hbb_r50_fpn_1x_dota_oc_20220121_095315.log.json) |
| ResNet50 (1024,1024,200) | 69.94 | oc | 1x | 3.39 | 15.6 | - | 2 | [rotated_retinanet_hbb_kld_r50_fpn_1x_dota_oc](./rotated_retinanet_hbb_kld_r50_fpn_1x_dota_oc.py) | [model](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_hbb_kld_r50_fpn_1x_dota_oc/rotated_retinanet_hbb_kld_r50_fpn_1x_dota_oc-49c1f937.pth) \| [log](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_hbb_kld_r50_fpn_1x_dota_oc/rotated_retinanet_hbb_kld_r50_fpn_1x_dota_oc_20220125_201832.log.json) |
| ResNet50 (1024,1024,200) | 69.86 | oc | 1x | 3.35 | 15.8 | - | 2 | [rotated_retinanet_hbb_kld_stable_r50_fpn_1x_dota_oc](./rotated_retinanet_hbb_kld_stable_r50_fpn_1x_dota_oc.py) | [model](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_hbb_kld_stable_r50_fpn_1x_dota_oc/rotated_retinanet_hbb_kld_stable_r50_fpn_1x_dota_oc-92a76443.pth) \| [log](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_hbb_kld_stable_r50_fpn_1x_dota_oc/rotated_retinanet_hbb_kld_stable_r50_fpn_1x_dota_oc_20220403_120631.log.json) |
| ResNet50 (1024,1024,200) | 68.42 | le90 | 1x | 3.38 | 16.9 | - | 2 | [rotated_retinanet_obb_r50_fpn_1x_dota_le90](../rotated_retinanet/rotated_retinanet_obb_r50_fpn_1x_dota_le90.py) | [model](https://download.openmmlab.com/mmrotate/v0.1.0/rotated_retinanet/rotated_retinanet_obb_r50_fpn_1x_dota_le90/rotated_retinanet_obb_r50_fpn_1x_dota_le90-c0097bc4.pth) \| [log](https://download.openmmlab.com/mmrotate/v0.1.0/rotated_retinanet/rotated_retinanet_obb_r50_fpn_1x_dota_le90/rotated_retinanet_obb_r50_fpn_1x_dota_le90_20220128_130740.log.json) |
| ResNet50 (1024,1024,200) | 70.22 | le90 | 1x | 3.35 | 16.9 | - | 2 | [rotated_retinanet_obb_kld_stable_r50_fpn_1x_dota_le90](./rotated_retinanet_obb_kld_stable_r50_fpn_1x_dota_le90.py) | [model](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_obb_kld_stable_r50_fpn_1x_dota_le90/rotated_retinanet_obb_kld_stable_r50_fpn_1x_dota_le90-31193e00.pth) \| [log](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_obb_kld_stable_r50_fpn_1x_dota_le90/rotated_retinanet_obb_kld_stable_r50_fpn_1x_dota_le90_20220402_225531.log.json) |
| ResNet50 (1024,1024,200) | 71.30 | le90 | 1x | 3.61 | 16.9 | - | 2 | [rotated_retinanet_obb_kld_stable_r50_adamw_fpn_1x_dota_le90](./rotated_retinanet_obb_kld_stable_r50_adamw_fpn_1x_dota_le90.py) | [model](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_obb_kld_stable_r50_adamw_fpn_1x_dota_le90/rotated_retinanet_obb_kld_stable_r50_adamw_fpn_1x_dota_le90-474d9955.pth) \| [log](https://download.openmmlab.com/mmrotate/v0.1.0/kld/rotated_retinanet_obb_kld_stable_r50_adamw_fpn_1x_dota_le90/rotated_retinanet_obb_kld_stable_r50_adamw_fpn_1x_dota_le90_20220608_003758.log.json) |
| Backbone | mAP | Angle | lr schd | Mem (GB) | Inf Time (fps) | Aug | Batch Size | Configs | Download |
| :----------------------: | :---: | :---: | :-----: | :------: | :------------: | :-: | :--------: | :-----------------------------------------------------------------------------: | :--------------------------------------------------------------------------------------------------------------
没有合适的资源?快使用搜索试试~ 我知道了~
df竞赛&毕设实验一.zip
共450个文件
py:287个
md:55个
yml:24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
0 下载量 19 浏览量
2023-10-20
21:03:22
上传
评论
收藏 11.7MB ZIP 举报
温馨提示
df竞赛&毕设实验一.zip
资源推荐
资源详情
资源评论
收起资源包目录
df竞赛&毕设实验一.zip (450个子文件)
make.bat 760B
make.bat 760B
CITATION.cff 282B
setup.cfg 509B
readthedocs.css 143B
readthedocs.css 143B
Dockerfile 1KB
Dockerfile 1KB
Dockerfile 527B
.gitattributes 66B
.gitignore 1KB
.gitignore 182B
mmrotate.iml 651B
MANIFEST.in 164B
MMRotate_Tutorial.ipynb 17.08MB
demo.jpg 399KB
zhihu_qrcode.jpg 388KB
zhihu_qrcode.jpg 388KB
zhihu_qrcode.jpg 388KB
result.jpg 387KB
dota_demo.jpg 275KB
qq_group_qrcode.jpg 200KB
qq_group_qrcode.jpg 200KB
qq_group_qrcode.jpg 200KB
class_indices.json 11KB
ms_trainval.json 598B
ms_train.json 473B
ms_val.json 467B
ms_trainval.json 462B
ss_trainval.json 460B
ss_trainval.json 454B
ms_test.json 403B
ms_extra.json 401B
ms_train.json 399B
ss_val.json 395B
ms_val.json 393B
ss_extra.json 388B
ss_train.json 388B
ss_train.json 386B
ss_test.json 385B
ss_val.json 380B
ms_test.json 364B
ss_test.json 346B
LICENSE 1KB
Makefile 634B
Makefile 634B
model_zoo.md 31KB
model_zoo.md 30KB
customize_config.md 18KB
customize_config.md 18KB
README.md 15KB
customize_runtime.md 12KB
README_zh-CN.md 12KB
customize_runtime.md 12KB
README_EN.md 11KB
README.md 9KB
customize_models.md 9KB
useful_tools.md 9KB
customize_models.md 9KB
intro.md 9KB
useful_tools.md 9KB
faq.md 9KB
README.md 8KB
faq.md 8KB
customize_dataset.md 8KB
customize_dataset.md 8KB
changelog.md 8KB
intro.md 8KB
README.md 7KB
README.md 7KB
install.md 7KB
install.md 6KB
README.md 6KB
README.md 6KB
README.md 6KB
README.md 5KB
get_started.md 5KB
get_started.md 5KB
README.md 5KB
README.md 5KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 4KB
README.md 3KB
README.md 3KB
README.md 2KB
command.md 2KB
README.md 2KB
README.md 2KB
README.md 1KB
README.md 1020B
README.md 829B
README.md 757B
README.md 435B
switch_language.md 145B
switch_language.md 145B
ss.md 37B
共 450 条
- 1
- 2
- 3
- 4
- 5
资源评论
天天501
- 粉丝: 606
- 资源: 4665
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