Inflated Episodic Memory With Region Self-Attention for Long-Tailed Visual Recognition

There have been increasing interests in modeling longtailed data. Unlike artificially collected datasets, longtailed data are naturally existed in the real-world and thus more realistic. To deal with the class imbalance problem, we introduce an Inflated Episodic Memory (IEM) for longtailed visual recognition. First, our IEM augments the convolutional neural networks with categorical representative features for rapid learning on tail classes. In traditional few-shot learning, a single prototype is usually leveraged to represent a category. However, long-tailed data has higher intra-class variances. It could be challenging to learn a single prototype for one category. Thus, we introduce IEM to store the most discriminative feature for each category individually. Besides, the memory banks are updated independently, which further decreases the chance of learning skewed classifiers. Second, we introduce a novel region self-attention mechanism for multi-scale spatial feature map encoding. It is beneficial to incorporate more discriminative features to improve generalization on tail classes. We propose to encode local feature maps at multiple scales, and the spatial contextual information should be aggregated at the same time. Equipped with IEM and region self-attention, we achieve state-of-the-art performance on four standard long-tailed image recognition benchmarks. Besides, we validate the effectiveness of IEM on a longtailed video recognition benchmark, i.e., YouTube-8M.