Average mAP: 29.9mAP IOU@0.1: 49.1mAP IOU@0.2: 47.5mAP IOU@0.3: 44.3
Papers with Code · CC BY-SA 4.0
Zero-Shot Learning
Zero-shot learning (ZSL) is a model's ability to detect classes never seen during training. The condition is that the classes are not known during supervised learning.
Earlier work in zero-shot learning use attributes in a two-step approach to infer unknown classes. In the computer vision context, more recent advances learn mappings from image feature space to semantic space. Other approaches learn non-linear multimodal embeddings. In the modern NLP context, language models can be evaluated on downstream tasks without fine tuning.
Benchmark datasets for zero-shot learning include aPY, AwA, and CUB, among others.
( Image credit: Prototypical Networks for Few shot Learning in PyTorch )
Further readings:
논문
mAP IOU@0.5: 32.2
Recall: 31.6
mAP IOU@0.5: 40.2
Average mAP: 37.5mAP IOU@0.2: 57.7mAP IOU@0.5: 42.7mAP IOU@0.7: 24.3
Avg mAP (0.1-0.5): 23.5mAP IOU@0.1: 26.6mAP IOU@0.2: 25.4mAP IOU@0.3: 24.2
Avg mAP (0.3:0.7): 59.6
mAP IOU@0.5: 23.3mAP IOU@0.3: 40.1mAP IOU@0.4: 29.4mAP IOU@0.6: 13.1
Average mAP: 32.64mAP IOU@0.1: 53.42mAP IOU@0.3: 47.41mAP IOU@0.5: 35.18
Avg mAP (0.1-0.5): 29.3mAP IOU@0.1: 33.1mAP IOU@0.2: 32.2mAP IOU@0.3: 30.4
mAP IOU@0.5: 57.18
Recall: 33.6
Average mAP: 14.2
Average mAP: 16.2
mAP@0.3: 25.9mAP@0.4: 22.6mAP@0.5: 18.9mAP@0.6: 15.0
Average mAP: 17.3
Average mAP: 23.5mAP IOU@0.1: 42.3mAP IOU@0.2: 39.7 mAP IOU@0.3: 35.8
Recall: 42.5
Avg mAP (0.1-0.5): 25.4mAP IOU@0.1: 28.6mAP IOU@0.2: 27.4mAP IOU@0.3: 26.1
mAP@0.2: 61.8
Recall: 13.3
Recall: 47.3
Recall: 46.5
mAP IOU@0.5: 35.2mAP IOU@0.1: 60.5mAP IOU@0.3: 48.4mAP IOU@0.7: 16.3