结果汇总：
| 数据库 | SUN RGB-D | NYU V1 | NYU V2 | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 方法 | Pixel | Mean | mIoU | Pixel | Mean | mIoU | Pixel | Mean | mIoU | f.w. Iou | 期刊 | 时间 | 备注 | |||||
| RedNet[1] | 81.3 % | 60.3% | 47.8% | - | - | - | - | - | - | arxiv | 2018 | |||||||
| Multimodal-RNNs[2] | - | - | - | 78.89% | 75.73% | 65.70% | 67.90% | 54.67% | 43.27% | arxiv | 2018 | |||||||
| S-M Fusion[3] | 78.07% | 53.93% | 40.98% | - | - | - | - | - | - | ICIP | 2018 | |||||||
| LSDNGF[4] | - | - | - | - | - | - | 71.9% | 60.7% | 45.9% | 59.3 % | cvpr | 2017 | ||||||
参考文献
-
Residual Encoder-Decoder Network for indoor RGB-D Semantic Segmentation, https://arxiv.org/abs/1806.01054 ↩
-
Multimodal Recurrent Neural Networks with Information Transfer Layers for Indoor Scene Labeling, https://arxiv.org/abs/1803.04687 ↩
-
SEMANTICS-GUIDED MULTI-LEVEL RGB-D FEATURE FUSION FOR INDOOR SEMANTIC,https://ieeexplore.ieee.org/iel7/8267582/8296222/08296484.pdf ↩
-
Locality-Sensitive Deconvolution Networks with Gated Fusion for RGB-D Indoor Semantic Segmentation,http://openaccess.thecvf.com/content_cvpr_2017/papers/Cheng_Locality-Sensitive_Deconvolution_Networks_CVPR_2017_paper.pdf ↩
网友评论