尚未阅读的各类文章
# 基于 CAM 擦除的
1、Adversarial Complementary Learning for Weakly Supervised ObjectLocalization
2、Object Region Mining with Adversarial Erasing: A Simple Classification to Semantic Segmentation Approach
3、Erasing Integrated Learning : A Simple yet Effective Approach for Weakly Supervised
# 基于图像增强的
1、Puzzle-CAM: Improved localization via matching partial and full features.
2、Self-supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentaion.
# 生成 CAM 的示意图
[ClassActivationMaps_PyTorch (opens new window)](https://github.com/maubreville/ClassActivationMaps_PyTorch/blob/master/ClassActivationMaps_Demo_Resnet18.ipynb)
跨越时空的难样本挖掘：https://www.zhihu.com/collection/148739489
优Tech分享 | 腾讯优图在弱监督目标定位的研究及应用：https://zhuanlan.zhihu.com/p/393262933
语义分割模型架构演进与相关论文阅读：https://blog.csdn.net/kevin_zhao_zl/article/details/106910045
图像语义分割(12)-重新思考空洞卷积: 为弱监督和半监督语义分割设计的简捷方法：https://zhuanlan.zhihu.com/p/52935388
(NeurIPS 2019) Gated CRF Loss-一种用于弱监督图像语义分割的新型损失函数：https://zhuanlan.zhihu.com/p/83964531
(CVPR2019)图像语义分割(22)FickleNet-随机推理用于弱监督和半监督图像语义分割：https://zhuanlan.zhihu.com/p/81707287
- 02
- README 美化05-20
- 03
- 常见 Tricks 代码片段05-12