Object Detection on LVIS all categories v1.0 (minival5k train)
60.5AP (All Categories)DetCLIPv3
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DetCLIPv3Backbone=Swin-L, Fine-tuning protocol=all categories2024.04 | 60.5 | 60.7 | |
| DetCLIPv2Backbone=Swin-L, Fine-tuning protocol=all categories2024.04 | 60.1 | 58.3 | |
| GLIPv2Backbone=Swin-H, Fine-tuning protocol=all categories2024.04 | 59.8 | — | |
| DetCLIPv3Backbone=Swin-T, Fine-tuning protocol=all categories2024.04 | 56.5 | 55.1 | |
| DetCLIPv2Backbone=Swin-T, Fine-tuning protocol=all categories2024.04 | 50.7 | 44.3 | |
| GLIPv2Backbone=Swin-T, Fine-tuning protocol=all categories2024.04 | 50.6 | — |