Search Results for author:
Found 2 papers, 1 papers with code
Multimodal Noisy Segmentation based fragmented burn scars identification in Amazon Rainforest
However, the task to segment burn marks is difficult because of its indistinguishably with similar looking land patterns, severe fragmented nature of burn marks and partially labelled noisy datasets.
Teaching CNNs to mimic Human Visual Cognitive Process & regularise Texture-Shape bias
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.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
