Skip-Clip: Self-Supervised Spatiotemporal Representation Learning by Future Clip Order Ranking

28 Oct 2019  ·  Alaaeldin El-Nouby, Shuangfei Zhai, Graham W. Taylor, Joshua M. Susskind ·

Deep neural networks require collecting and annotating large amounts of data to train successfully. In order to alleviate the annotation bottleneck, we propose a novel self-supervised representation learning approach for spatiotemporal features extracted from videos. We introduce Skip-Clip, a method that utilizes temporal coherence in videos, by training a deep model for future clip order ranking conditioned on a context clip as a surrogate objective for video future prediction. We show that features learned using our method are generalizable and transfer strongly to downstream tasks. For action recognition on the UCF101 dataset, we obtain 51.8% improvement over random initialization and outperform models initialized using inflated ImageNet parameters. Skip-Clip also achieves results competitive with state-of-the-art self-supervision methods.

PDF Abstract


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Self-Supervised Action Recognition UCF101 Skip-Clip (3D ResNet-18) 3-fold Accuracy 64.4 # 44
Pre-Training Dataset UCF101 # 1
Frozen false # 1