Teaching CNNs to mimic Human Visual Cognitive Process & regularise Texture-Shape bias
This work addresses the texture-shape bias in CNNs for computer vision researchers, offering an incremental improvement with enhanced explainability.
The paper tackled the problem of CNNs' over-reliance on texture over shape for object recognition by proposing CognitiveCNN, an architecture inspired by human visual cognition that uses interpretable features like shape and texture, and introduced a regularization method to balance their influence, resulting in improved accuracy and robustness.
Recent experiments in computer vision demonstrate texture bias as the primary reason for supreme results in models employing Convolutional Neural Networks (CNNs), conflicting with early works claiming that these networks identify objects using shape. It is believed that the cost function forces the CNN to take a greedy approach and develop a proclivity for local information like texture to increase accuracy, thus failing to explore any global statistics. We propose CognitiveCNN, a new intuitive architecture, inspired from feature integration theory in psychology to utilise human interpretable feature like shape, texture, edges etc. to reconstruct, and classify the image. We define novel metrics to quantify the "relevance" of "abstract information" present in these modalities using attention maps. We further introduce a regularisation method which ensures that each modality like shape, texture etc. gets proportionate influence in a given task, as it does for reconstruction; and perform experiments to show the resulting boost in accuracy and robustness, besides imparting explainability to these CNNs for achieving superior performance in object recognition.