SenseTime open-sources SenseNova-Vision unified vision model

The model handles object detection, OCR, keypoint localization, segmentation, depth estimation, and 3D reconstruction. It is fully open-sourced as part of the SenseNova foundation-model suite.