🧐7 things you should know about the Focal Loss: 📌 It was introduced in the RetinaNet paper to address the foreground-background class imbalance encountered during training of dense detectors (one-stage detectors) ...

🥇YOLOX beat YOLOv5. They won the 1st Place on Streaming Perception Challenge (Workshop on Autonomous Driving at CVPR 2021). I briefly explain how they created it, here below👇 YOLOX will be soon supported in our IceVision Framework (Repo: github.com/airctic/IceVision)

#SmallObjectDetection Here is another good technique to detect small objects when you have access to high resolution images. Inference is done on patches/slices and then stitched together. Repo: github.com/obss/sahi

#SmallObjectDetection Interesting approach for detecting tiny objects as small as 2 × 2 pixels. It uses Multi-Scale Semantic Segmentation. Anchor free. Paper: Oriented Bounding Boxes for Small and Freely Rotated Objects pdf: arxiv.org/pdf/2104.11854.pdf abs: arxiv.org/abs/2104.11854

🎁 Small-Objects Detection Tips & Tricks 🎁 - Anchor Boxes Use anchor boxes with size/ratio close to target boxes - Image Size • Use the highest image resolution your GPU can afford. • Use progressive resizing: Train with small size image, and gradually increase size 🧵