Ultralytics' YOLOv8.1 brings major enhancements to its popular object recognition framework

Ultralytics has released YOLOv8.1, a major update to its object recognition framework. The new version introduces Oriented Bounding Box (OBB) models, which improve accuracy and reduce background noise for angled or rotated objects in applications such as aerial imagery and text recognition. YOLOv8.1 also includes enhanced segmentation capabilities, performance optimizations, an improved model architecture, an improved API and CLI experience, and a more intuitive user interface with the Ultralytics Explorer for exploring datasets. In addition, the update adds support for multiple platforms, including PaddlePaddle, NCNN, PNNX, and TensorRT. According to Ultralytics, 19 million YOLOv8 models have been trained in 2023, with more than 20 million downloads, resulting in 1,000 inference jobs per second every day.