Object Detection and Segmentation

• Object detection problem: bounding boxes, IoU, mAP
• Two-stage detectors: R-CNN, Fast R-CNN, Faster R-CNN (region proposal networks), anchor boxes
• One-stage detectors: YOLO family, SSD, RetinaNet (focal loss)
• Anchor-free detection: FCOS, CenterNet, CornerNet
• Semantic segmentation: FCN, U-Net (skip connections), DeepLab (atrous/dilated convolutions, CRF)
• Instance segmentation: Mask R-CNN
• Panoptic segmentation: unifying semantic and instance segmentation
• Real-time segmentation: BiSeNet, DDRNet