Counterfactual Attention Learning for Fine-Grained Visual Categorization and Re-identification

Benchmark Model Rank Results
few-shot-learning-on-dtdCAL#34-shot Accuracy: 40.98-shot Accuracy: 50.412-shot Accuracy: 54.6
few-shot-learning-on-fgvc-aircraft-1CAL#2Harmonic mean: 35.24-shot Accuracy: 35.28-shot Accuracy: 55.4
few-shot-learning-on-stanford-carsCAL#34-shot Accuracy: 42.28-shot Accuracy: 71.812-shot Accuracy: 82.9
fine-grained-image-classification-on-cub-200-1CAL#8Accuracy: 90.6
fine-grained-image-classification-on-fgvcCAL#9Accuracy: 94.2
fine-grained-image-classification-on-stanfordCAL#10Accuracy: 95.5%
mitigating-contextual-bias-on-fgvc-aircraftCAL + ALIA#2Top-1 Accuracy (%): 71.8OOD Accuracy (%): 25.1
mitigating-contextual-bias-on-fgvc-aircraftCAL#4Top-1 Accuracy (%): 71.0OOD Accuracy (%): 10.2
person-re-identification-on-dukemtmc-reidCAL#40mAP: 80.5Rank-1: 90
person-re-identification-on-market-1501CAL#47Rank-1: 95.5mAP: 89.5
person-re-identification-on-msmt17CAL(ResNet50)#21mAP: 64Rank-1: 84.2
vehicle-re-identification-on-vehicleid-largeCAL#8Rank-1: 75.1mAP: 80.9
vehicle-re-identification-on-vehicleid-mediumCAL#7Rank-1: 78.2mAP: 83.8
vehicle-re-identification-on-vehicleid-smallCAL#9Rank-1: 82.5mAP: 87.8
vehicle-re-identification-on-veri-776CAL#12mAP: 74.3Rank-1: 95.4Rank5: 97.9