Co-segmentation Inspired Attention Module for Video-based Computer Vision Tasks
This work addresses the need for more efficient and less biased attention mechanisms in video analysis, offering a generic module for various tasks, though it is incremental as it builds on existing co-segmentation and attention concepts.
The paper tackles the problem of improving video-based computer vision tasks by automatically focusing on salient regions without relying on pre-trained models, proposing a Co-Segmentation inspired Attention Module (COSAM) that leads to notable performance improvements in tasks like person re-ID, video captioning, and action classification.
Video-based computer vision tasks can benefit from estimation of the salient regions and interactions between those regions. Traditionally, this has been done by identifying the object regions in the images by utilizing pre-trained models to perform object detection, object segmentation and/or object pose estimation. Although using pre-trained models is a viable approach, it has several limitations in the need for an exhaustive annotation of object categories, a possible domain gap between datasets, and a bias that is typically present in pre-trained models. In this work, we propose to utilize the common rationale that a sequence of video frames capture a set of common objects and interactions between them, thus a notion of co-segmentation between the video frame features may equip the model with the ability to automatically focus on task-specific salient regions and improve the underlying task's performance in an end-to-end manner. In this regard, we propose a generic module called ``Co-Segmentation inspired Attention Module'' (COSAM) that can be plugged in to any CNN model to promote the notion of co-segmentation based attention among a sequence of video frame features. We show the application of COSAM in three video-based tasks namely: 1) Video-based person re-ID, 2) Video captioning, & 3) Video action classification and demonstrate that COSAM is able to capture the task-specific salient regions in video frames, thus leading to notable performance improvements along with interpretable attention maps for a variety of video-based vision tasks, with possible application to other video-based vision tasks as well.