Instance Segmentation on ARMBench
AP50
- 2023-03-29 — Mask R-CNN (Resnet50): AP50 72.0
- 2023-10-16 — RoboLLM (VIT-B): AP50 82.0
- 2024-07-01 — RISE (VIT-B): AP50 86.37
| Rank | Model | AP50 | AP75 | Paper | Code | Year |
|---|---|---|---|---|---|---|
| 1 | RISE (VIT-B) | 86.37 | 77.51 | Robot Instance Segmentation with Few Annotations for Grasping | mkimhi/RISE | 2024 |
| 2 | RISE (R101) | 84.74 | 75.93 | Robot Instance Segmentation with Few Annotations for Grasping | mkimhi/RISE | 2024 |
| 3 | RISE (R50) | 83.53 | 75.15 | Robot Instance Segmentation with Few Annotations for Grasping | mkimhi/RISE | 2024 |
| 4 | RoboLLM (VIT-B) | 82.0 | 74 | RoboLLM: Robotic Vision Tasks Grounded on Multimodal Large Language Models | longkukuhi/armbench | 2023 |
| 5 | Mask2Former | 81.2 | 74.0 | Robot Instance Segmentation with Few Annotations for Grasping | mkimhi/RISE | 2024 |
| 6 | Deformable DETR | 77.03 | 63.4 | Robot Instance Segmentation with Few Annotations for Grasping | mkimhi/RISE | 2024 |
| 7 | Mask R-CNN (Resnet50) | 72 | 61 | ARMBench: An Object-centric Benchmark Dataset for Robotic Manipulation | 2023 |