Description
Sample Consistency Network (SCNet) is a method for instance segmentation which ensures the IoU distribution of the samples at training time are as close to that at inference time. To this end, only the outputs of the last box stage are used for mask predictions at both training and inference. The Figure shows the IoU distribution of the samples going to the mask branch at training time with/without sample consistency compared to that at inference time.
Papers Using This Method
Multiple Information Prompt Learning for Cloth-Changing Person Re-Identification2024-11-01SCNet: Sparse Compression Network for Music Source Separation2024-01-24Semantic-aware Consistency Network for Cloth-changing Person Re-Identification2023-08-27Self-Supervised Scalable Deep Compressed Sensing2023-08-26Underwater Image Enhancement via Learning Water Type Desensitized Representations2021-02-01SCNet: Training Inference Sample Consistency for Instance Segmentation2020-12-18