Object Identification on ARMBench (test)
98Recall@1RoboLLM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RoboLLMRef Set=Container, N=3, ARMBench fine-tuning=true2023.10 | 98 | 98.1 | 98.2 | |
| RoboLLMRef Set=Container, N=1, ARMBench fine-tuning=true2023.10 | 97.8 | 97.9 | 98 | |
| BEIT-3-BaseRef Set=Container, N=3, ARMBench fine-tuning=false2023.10 | 84.5 | — | — | |
| BEIT-3-BaseRef Set=Container, N=1, ARMBench fine-tuning=false2023.10 | 83.7 | 83.8 | 84.5 | |
| DINO-VIT-SRef Set=Container, N=3, ARMBench fine-tuning=true2023.10 | 79.5 | 89.4 | 93.5 | |
| RoboLLMRef Set=All refs, N=3, ARMBench fine-tuning=true2023.10 | 78.2 | 85.7 | 89.1 | |
| DINO-VIT-SRef Set=Container, N=1, ARMBench fine-tuning=true2023.10 | 77.2 | 87.3 | 91.6 | |
| RoboLLMRef Set=All refs, N=1, ARMBench fine-tuning=true2023.10 | 74.6 | 82.6 | 85.3 | |
| ResNet50-RMACRef Set=Container, N=3, ARMBench fine-tuning=true2023.10 | 72.2 | 82.9 | 88.2 | |
| ResNet50-RMACRef Set=Container, N=1, ARMBench fine-tuning=true2023.10 | 71.7 | 81.9 | 87.2 |