AR@100: 50.26AR@1000: 68.19AR@200: 57.30AR@50: 44.56
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
AR@100: 50.86AR@1000: 67.34AR@200: 57.56AR@50: 64.28
mAP@0.5: 30.3
mAP@0.5: 37.2Mean mAP: 21.7
Average Precision: 48.23Average Recall: 80.31Average F1: 60.27
AUC (val): 69.16AR@100: 76.72AUC (test): 69.26
mAP@0.5: 49.6Mean mAP: 29.9
AEI: Actors-Environment Interaction with Adaptive Attention for Temporal Action Proposals Generation
AUC (val): 69.47AR@100: 77.24AUC (test): 70.09
mAP@0.5: 27.3
mAP@0.5: 38.5
AR@100: 46.06AR@1000: 64.52AR@200: 53.21AR@50: 37.46
mAP@0.5: 42.7Mean mAP: 26.1
mAP@0.5: 36.8
D2-Net: Weakly-Supervised Action Localization via Discriminative Embeddings and Denoised Activations
mAP@0.5: 42.3Mean mAP: 26
mAP@0.5: 44Mean mAP: 26.8
mAP@0.5: 33.7
mAP@0.5: 48.5Mean mAP: 30.8
mAP@0.5: 44.2Mean mAP: 26.5
mAP@0.5: 43.8Mean mAP: 27.9
mAP@0.5: 37.8Mean mAP: 22.8
mAP@0.5: 37.0
mAP@0.5: 29.3
mAP@0.5: 40.2Mean mAP: 25.8
mAP@0.5: 37.1
mAP@0.5: 48.6Mean mAP: 29.9
mAP@0.5: 41.0Mean mAP: 24.4
mAP@0.5: 41.2Mean mAP: 25.9