Depth Estimation on KITTI depth (val)
0.975Acc (δ < 1.25)DepthFormer
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DepthFormerBackbone=ResNet-50 + Swin-L, Pre-trained on ImageNet-22K=true2023.03 | 0.975 | 0.997 | 0.999 | 0.052 | 0.158 | 2.143 | 0.079 | — | |
| DDPBackbone=Swin-L, Pre-trained on ImageNet-22K=true, Diffusion steps=32023.03 | 0.975 | 0.997 | 0.999 | 0.05 | 0.148 | 2.072 | 0.076 | — | |
| BinsFormerBackbone=Swin-L, Pre-trained on ImageNet-22K=true2023.03 | 0.974 | 0.997 | 0.999 | 0.052 | 0.151 | 2.098 | 0.079 | — | |
| DDPBackbone=Swin-B, Pre-trained on ImageNet-22K=true, Diffusion steps=32023.03 | 0.973 | 0.997 | 0.999 | 0.051 | 0.155 | 2.119 | 0.078 | — | |
| DDPBackbone=Swin-S, Diffusion steps=32023.03 | 0.97 | 0.996 | 0.999 | 0.053 | 0.167 | 2.171 | 0.082 | — | |
| DDPBackbone=Swin-T, Diffusion steps=32023.03 | 0.969 | 0.996 | 0.999 | 0.054 | 0.168 | 2.172 | 0.083 | — | |
| DepthFormerBackbone=ResNet-50 + Swin-T2023.03 | 0.966 | 0.995 | 0.999 | 0.056 | 0.177 | 2.252 | 0.086 | — | |
| AdaBinsBackbone=EfficientNet-B5 + Mini-ViT2023.03 | 0.964 | 0.995 | 0.999 | 0.058 | 0.19 | 2.36 | 0.088 | — | |
| ViP-DeepLab2020.12 | 0.9627 | 0.9941 | 0.9981 | 5.72 | 0.96 | 2.58 | 0.092 | 8.47 | |
| DPTBackbone=ResNet-50 + ViT-B2023.03 | 0.959 | 0.995 | 0.999 | 0.062 | — | 2.573 | 0.092 | — | |
| [20]2020.12 | 0.9577 | 0.9921 | 0.9975 | 6.99 | 1.27 | 2.86 | 0.104 | 9.73 | |
| BTSBackbone=DenseNet-1612023.03 | 0.956 | 0.993 | 0.998 | 0.059 | 0.245 | 2.756 | 0.096 | — | |
| TransDepthBackbone=ResNet-50 + ViT-B2023.03 | 0.956 | 0.994 | 0.999 | 0.064 | 0.252 | 2.755 | 0.098 | — | |
| DepthGenBackbone=Efficient U-Net, Diffusion steps=82023.03 | 0.953 | 0.991 | 0.998 | 0.064 | 0.356 | 2.985 | 0.1 | — | |
| VNLBackbone=ResNeXt-1012023.03 | 0.938 | 0.99 | 0.998 | 0.072 | — | 3.258 | 0.117 | — | |
| DORNBackbone=ResNet-1012023.03 | 0.932 | 0.984 | 0.994 | 0.072 | 0.307 | 2.727 | 0.12 | — |