Accuracy: 98.02%
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: 41.0mAP IOU@0.75: 24.9mAP IOU@0.95: 6.2mAP: 25.1
mAP IOU@0.1: 70.1mAP IOU@0.2: 63.6mAP IOU@0.3: 54.5mAP IOU@0.4: 45.7
Accuracy: 97.45
Cross Subject Accuracy: 88.8Cross View Accuracy: 96.3
10%: 4
mAP IOU@0.1: 61.2mAP IOU@0.2: 56.1mAP IOU@0.3: 48.1mAP IOU@0.4: 39.0
Actions Top-1: 28.03
Cross Subject Accuracy: 91.4Cross View Accuracy: 96.7
Actions Top-1: 28.7Verbs Top-1: 65.7Object Top-1: 36.3
Accuracy: 91.7%
PCK@0.1: 45.2PCK@0.2: 62.9PCK@0.3: 73.5PCK@0.4: 80.6
Actions Top-1: 28.07Verbs Top-1: 62.66Object Top-1: 31.69
PCK@0.1: 58.4PCK@0.2: 78.1PCK@0.3: 85.9PCK@0.4: 89.8
Cross Subject Accuracy: 92.3Cross View Accuracy: 96.4
Actions Top-1: 41.06Verbs Top-1: 69.23Object Top-1: 51.17
Accuracy: 98.4%
Cross Subject Accuracy: 90.2Cross View Accuracy: 96.4
Cross Subject Accuracy: 90.5Cross View Accuracy: 96.5
Actions Top-1: 33.61Verbs Top-1: 61.99Object Top-1: 42.90
Accuracy: 91.40%
Accuracy: 99.1%
PCK@0.1: 45.2PCK@0.2: 69.6PCK@0.3: 80.8PCK@0.4: 87.5
Actions Top-1: 35.27Verbs Top-1: 58.27Object Top-1: 47.45
Actions Top-1: 26.7Verbs Top-1: 64.4Object Top-1: 33.8