Image Classification on PASCAL VOC 2007 (mAP)
92.5mAPMLCD
Evaluation results
| Method | Links | |
|---|---|---|
| MLCDPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 92.5 | |
| OpenCLIPPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 91.7 | |
| UNICOMPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 91.4 | |
| CLIP+Pre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 89.7 | |
| CLIPPre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.07 | 89.6 | |
| SimCLR ResNet50x4training dataset=ImageNet2020.06 | 84.2 | |
| SimCLR ResNet50training dataset=ImageNet2020.06 | 80.5 | |
| MMV TSMx2image handling strategy=deflation, training dataset=AS+HT2020.06 | 77.4 | |
| MMV TSMx2image handling strategy=input-inflation, training dataset=AS+HT2020.06 | 77.4 | |
| MMV TSMimage handling strategy=input-inflation, training dataset=AS+HT2020.06 | 75.7 | |
| MMV TSMimage handling strategy=deflation, training dataset=AS+HT2020.06 | 74.8 | |
| MMV S3D-Gimage handling strategy=input-inflation, training dataset=AS+HT2020.06 | 72.1 | |
| MMV S3D-Gimage handling strategy=deflation, training dataset=AS+HT2020.06 | 71.4 | |
| Supervised S3D-Gimage handling strategy=deflation, training dataset=Kinetics2020.06 | 67.9 | |
| Supervised TSMimage handling strategy=deflation, training dataset=Kinetics2020.06 | 66.9 | |
| Supervised TSMx2image handling strategy=deflation, training dataset=Kinetics2020.06 | 66.9 | |
| MMV TSMx2image handling strategy=naive deflation, training dataset=AS+HT2020.06 | 45.6 | |
| MMV S3D-Gimage handling strategy=naive deflation, training dataset=AS+HT2020.06 | 41.8 | |
| MMV TSMimage handling strategy=naive deflation, training dataset=AS+HT2020.06 | 34.4 |