Detectron2 - Next Gen Object Detection Library
summary
Detectron2 is a ground-up PyTorch rewrite of Detectron that implements Mask R-CNN, panoptic and instance segmentation, TensorMask and other models, claims 2–3x faster training than Detectron1 and fixes implementation quirks that yield measurable accuracy gains (historically ~1 point, with further improvement reported), includes large pre-trained models and smaller optimized variants for production, and already runs in Facebook servers, on mobile and Portal devices to power detection and segmentation at scale.
AI-written summary · about
chapters (4)
related talks
28mObject Detection using Transformers and CNNs
36mObject detection to Instance segmentation: Learn to apply several algorithms...
22mObject Detection with KerasCV
38mDeep Learning for detection on a phone
01h 15mTrain Object Detection with Small Datasets
12mCascade models in computer vision to boost accuracy and performance
59mDeep Neural Networks for Video Applications
13mEnhancing Object Detection & Avoidance Systems for Robots in Challenging Environments Using AI