MILA: Multi-Task Learning from Videos via Efficient Inter-Frame Attention

Prior work in multi-task learning has mainly focused on predictions on a single image. In this work, we present a new approach for multi-task learning from videos via efficient inter-frame local attention (MILA). Our approach contains a novel inter-frame attention module which allows learning of task-specific attention across frames. We embed the attention module in a "slow-fast" architecture, where the slow network runs on sparsely sampled keyframes and the fast shallow network runs on non-keyframes at a high frame rate. We also propose an effective adversarial learning strategy to encourage the slow and fast net-work to learn similar features to well align keyframes and non-keyframes. Our approach ensures low-latency multi-task learning while maintaining high quality predictions. MILA obatins competitive accuracy compared to state-of-the-art on two multi-task learning benchmarks while reducing the number of floating point operations (FLOPs) by up to 70%. In addition, our attention based feature propagation method (ILA) outperforms prior work in terms of task accuracy while also reducing up to 90% of FLOPs.