Instance Segmentation on Habitat Gibson Specialization 1.0 (val)
37.28Mask AP50SEAL
Evaluation results
| Method | Links | |
|---|---|---|
| SEALAction Policy=Gainful Curiosity, Perception Baseline=3DLabelProp2021.12 | 37.28 | |
| Semantic Curiosity + Optical FlowAction Policy=Semantic Curiosity, Perception Baseline=Optical Flow2021.12 | 33.29 | |
| Active Neural SLAM + Optical FlowAction Policy=Active Neural SLAM, Perception Baseline=Optical Flow2021.12 | 33.12 | |
| Semantic Curiosity + Self-trainingAction Policy=Semantic Curiosity, Perception Baseline=Self-training2021.12 | 32.88 | |
| Pretrained Mask-RCNNState=Pretrained2021.12 | 32.54 | |
| Frontier Exploration + Self-trainingAction Policy=Frontier Exploration, Perception Baseline=Self-training2021.12 | 32.5 | |
| Active Neural SLAM + Self-trainingAction Policy=Active Neural SLAM, Perception Baseline=Self-training2021.12 | 32.44 | |
| Random Policy + Optical FlowAction Policy=Random Policy, Perception Baseline=Optical Flow2021.12 | 32.22 | |
| Frontier Exploration + Optical FlowAction Policy=Frontier Exploration, Perception Baseline=Optical Flow2021.12 | 32.12 | |
| Random Policy + Self-trainingAction Policy=Random Policy, Perception Baseline=Self-training2021.12 | 31.23 |