最近在調研3D算法方面的工作,整理了幾篇多視角學習的文章。還沒調研完,先寫個大概。
基於RGBD的語義分割的工作重點主要集中在如何將RGB信息和Depth信息融合,主要分為三類:省略。
- 1、(ICCV2017)《RDFNet: RGB-D Multi-level Residual Feature Fusion for Indoor Semantic Segmentation》
- 2、(2018 Arxiv)RedNet:Residual Encoder-Decoder Network for indoor RGB-D Semantic Segmentation
- 3、(ICIP2019)ACNet:使用注意力網絡的RGBD圖像語義分割方法
- 4、(NIPS2020)Deep Multimodal Fusion by Channel Exchanging
- 5、(ECCV2020)Bi-directional Cross-Modality Feature Propagation with Separation-and-Aggregation Gate for RGB-D Semantic Segmentation
- 6、(arxiv2021)GLPNet:Global-Local Propagation Network for RGB-D Semantic Segmentation
- 7、(ACCV 2016) FuseNet: Incorporating Depth into Semantic Segmentation via Fusion-based CNN Architecture
- 8、(SCIA2017)Multimodal Neural Networks: RGB-D for Segmantic Segmentation and Object Detection
- 9、(3DV2019)3D Neighborhood Convolution: Learning Depth-Aware Features for RGB-D and RGB Semantic Segmentation
- 10、(ICCV2017)3D Graph Neural Networks for RGBD Semantic Segmentation
- 多模態Transformer
- Transformer語義分割(SETR)
- TransUNet:用於醫學圖像分割的Transformers強大編碼器
- SegFormer:使用Transformer進行語義分割的簡單高效設計
- Swin-Unet:首個純Transformer的醫學圖像分割網絡
- 學習跨模態深度表達用於多模態MR圖像分割
1、(ICCV2017)《RDFNet: RGB-D Multi-level Residual Feature Fusion for Indoor Semantic Segmentation》
用於室內語義分割的RGB-D多級殘差特征融合
論文地址:https://openaccess.thecvf.com/content_iccv_2017/html/Park_RDFNet_RGB-D_Multi-Level_ICCV_2017_paper.html
代碼:https://github.com/SeongjinPark/RDFNet
文章介紹:https://blog.csdn.net/u012113559/article/details/81363756
2、(2018 Arxiv)RedNet:Residual Encoder-Decoder Network for indoor RGB-D Semantic Segmentation
論文地址:https://arxiv.org/abs/1806.01054
代碼:https://github.com/JindongJiang/RedNet
文章介紹:https://blog.csdn.net/qq_41375318/article/details/104311597、
https://blog.csdn.net/qq_41375318/article/details/103451966
3、(ICIP2019)ACNet:使用注意力網絡的RGBD圖像語義分割方法
論文地址:https://arxiv.org/abs/1905.10089
代碼:https://github.com/anheidelonghu/ACNet
文章介紹:https://blog.csdn.net/kevin_zhao_zl/article/details/100750591、
https://zhuanlan.zhihu.com/p/82193530
4、(NIPS2020)Deep Multimodal Fusion by Channel Exchanging
論文地址:https://arxiv.org/abs/2011.05005
代碼:https://github.com/yikaiw/CEN
文章介紹:https://zhuanlan.zhihu.com/p/341959576、
https://blog.csdn.net/hongyuge/article/details/109632887
視頻講解:https://www.bilibili.com/video/BV1ya4y1W7Hf
5、(ECCV2020)Bi-directional Cross-Modality Feature Propagation with Separation-and-Aggregation Gate for RGB-D Semantic Segmentation
論文地址:https://arxiv.org/abs/2007.09183
代碼:https://github.com/charlesCXK/RGBD_Semantic_Segmentation_PyTorch
文章介紹:https://blog.csdn.net/sinat_17456165/article/details/107805136
6、(arxiv2021)GLPNet:Global-Local Propagation Network for RGB-D Semantic Segmentation
論文地址:https://arxiv.org/abs/2101.10801
代碼:無
文章介紹:
7、(ACCV 2016) FuseNet: Incorporating Depth into Semantic Segmentation via Fusion-based CNN Architecture
論文地址:https://www.semanticscholar.org/paper/FuseNet%3A-Incorporating-Depth-into-Semantic-via-CNN-Hazirbas-Ma/9360ce51ec055c05fd0384343792c58363383952
代碼:https://github.com/tum-vision/fusenet
文章介紹:https://blog.csdn.net/u013841196/article/details/82939619
8、(SCIA2017)Multimodal Neural Networks: RGB-D for Segmantic Segmentation and Object Detection
論文地址:https://www.researchgate.net/publication/317803469_Multimodal_Neural_Networks_RGB-D_for_Semantic_Segmentation_and_Object_Detection
代碼:
文章介紹:https://blog.csdn.net/qq_38316300/article/details/109546441
9、(3DV2019)3D Neighborhood Convolution: Learning Depth-Aware Features for RGB-D and RGB Semantic Segmentation
論文地址:https://arxiv.org/abs/1910.01460
代碼:
文章介紹:https://blog.csdn.net/cangafuture/article/details/113822865
10、(ICCV2017)3D Graph Neural Networks for RGBD Semantic Segmentation
論文地址:https://ieeexplore.ieee.org/document/8237818
代碼:https://github.com/yanx27/3DGNN_pytorch
文章介紹:https://blog.csdn.net/P_LarT/article/details/88774811、https://blog.csdn.net/P_LarT/article/details/88774811
多模態Transformer
論文地址:https://arxiv.org/abs/1906.00295
代碼:https://github.com/yaohungt/Multimodal-Transformer
論文介紹:https://zhuanlan.zhihu.com/p/84678022?from_voters_page=true、https://zhuanlan.zhihu.com/p/340113856、https://blog.csdn.net/zpainter/article/details/111867693
Transformer語義分割(SETR)
論文地址:https://arxiv.org/abs/2012.15840
代碼:https://github.com/fudan-zvg/SETR
文章介紹:https://zhuanlan.zhihu.com/p/341768446
TransUNet:用於醫學圖像分割的Transformers強大編碼器
論文地址:https://arxiv.org/abs/2102.04306
代碼:https://github.com/Beckschen/TransUNet
文章介紹:https://blog.csdn.net/weixin_49627776/article/details/115710379
SegFormer:使用Transformer進行語義分割的簡單高效設計
論文地址:https://arxiv.org/abs/2105.15203
代碼:https://github.com/NVlabs/SegFormer
文章介紹:https://zhuanlan.zhihu.com/p/379054782
Swin-Unet:首個純Transformer的醫學圖像分割網絡
論文地址:https://arxiv.org/abs/2105.05537
代碼:https://github.com/HuCaoFighting/Swin-Unet(目前未開源)
文章介紹:https://blog.csdn.net/amusi1994/article/details/116957208