Image Classification on PACS (ID/OOD)
99.11ID AccuracyCLIP-ZS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CLIP-ZSType=Training-free, Pre-trained model=YFCC100M ResNet50-CLIP, Training setup=joint training2024.03 | 99.11 | 49.12 | |
| Manually AnnotatedType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 97.03 | 33.8 | |
| SuS-X-SDType=Training-free, Pre-trained model=YFCC100M ResNet50-CLIP, Training setup=joint training2024.03 | 95.55 | 47.81 | |
| HIRPG + CONANType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 94.36 | 60.75 | |
| VisDescType=Training-free, Pre-trained model=YFCC100M ResNet50-CLIP, Training setup=joint training2024.03 | 93.77 | 46.09 | |
| LE + CONANType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 93.47 | 44.54 | |
| CHB + CONANType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 92.88 | 41.42 | |
| CALIPType=Training-free, Pre-trained model=YFCC100M ResNet50-CLIP, Training setup=joint training2024.03 | 92.58 | 48.43 | |
| SD-ClfType=Training-free, Pre-trained model=YFCC100M ResNet50-CLIP, Training setup=joint training2024.03 | 92.58 | 48.43 | |
| HIRPGType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 89.91 | 43.98 | |
| CuPLType=Training-free, Pre-trained model=YFCC100M ResNet50-CLIP, Training setup=joint training2024.03 | 89.32 | 46.51 | |
| LEType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 88.43 | 38.03 | |
| CCG + CONANType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 85.76 | 41.55 | |
| SC + CONANType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 85.46 | 42.05 | |
| Glide-synType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 85.16 | 33.2 | |
| CHBType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 83.38 | 30.98 | |
| CCGType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 81.01 | 31.71 | |
| SCType=Training-dependent, Pre-trained model=CLIP vision encoder, Training setup=joint training2024.03 | 76.26 | 28.19 |