Description
VarifocalNet is a method aimed at accurately ranking a huge number of candidate detections in object detection. It consists of a new loss function, named Varifocal Loss, for training a dense object detector to predict the IACS, and a new efficient star-shaped bounding box feature representation for estimating the IACS and refining coarse bounding boxes. Combining these two new components and a bounding box refinement branch, results in a dense object detector on the FCOS architecture, what the authors call VarifocalNet or VFNet for short.
Papers Using This Method
Class Imbalance Correction for Improved Universal Lesion Detection and Tagging in CT2025-04-08Universal Lymph Node Detection in Multiparametric MRI with Selective Augmentation2025-04-07Comparing Surface Landmine Object Detection Models on a New Drone Flyby Dataset2024-10-17How to Train an Accurate and Efficient Object Detection Model on Any Dataset2022-11-30Universal Lymph Node Detection in T2 MRI using Neural Networks2022-03-31Slicing Aided Hyper Inference and Fine-tuning for Small Object Detection2022-02-14SWA Object Detection2020-12-23VarifocalNet: An IoU-aware Dense Object Detector2020-08-31