Peripheral Vision Transformer
This work addresses visual recognition tasks by introducing a biologically inspired method, but it is incremental as it builds on existing transformer architectures with a novel encoding.
The authors tackled the problem of modeling human peripheral vision in deep neural networks for visual recognition by proposing PerViT, which incorporates peripheral position encoding into self-attention layers, resulting in performance improvements in image classification on ImageNet-1K across different model sizes.
Human vision possesses a special type of visual processing systems called peripheral vision. Partitioning the entire visual field into multiple contour regions based on the distance to the center of our gaze, the peripheral vision provides us the ability to perceive various visual features at different regions. In this work, we take a biologically inspired approach and explore to model peripheral vision in deep neural networks for visual recognition. We propose to incorporate peripheral position encoding to the multi-head self-attention layers to let the network learn to partition the visual field into diverse peripheral regions given training data. We evaluate the proposed network, dubbed PerViT, on ImageNet-1K and systematically investigate the inner workings of the model for machine perception, showing that the network learns to perceive visual data similarly to the way that human vision does. The performance improvements in image classification over the baselines across different model sizes demonstrate the efficacy of the proposed method.