Compact Generalized Non-local Network
This work addresses a specific bottleneck in computer vision for tasks like fine-grained recognition, but it is incremental as it builds upon existing non-local methods.
The paper tackles the limitation of non-local modules in capturing interactions between positions across channels, which is crucial for fine-grained object and action recognition, by generalizing the non-local module to account for correlations between positions of any two channels, resulting in clear-cut improvements on fine-grained object recognition and video classification tasks.
The non-local module is designed for capturing long-range spatio-temporal dependencies in images and videos. Although having shown excellent performance, it lacks the mechanism to model the interactions between positions across channels, which are of vital importance in recognizing fine-grained objects and actions. To address this limitation, we generalize the non-local module and take the correlations between the positions of any two channels into account. This extension utilizes the compact representation for multiple kernel functions with Taylor expansion that makes the generalized non-local module in a fast and low-complexity computation flow. Moreover, we implement our generalized non-local method within channel groups to ease the optimization. Experimental results illustrate the clear-cut improvements and practical applicability of the generalized non-local module on both fine-grained object recognition and video classification. Code is available at: https://github.com/KaiyuYue/cgnl-network.pytorch.