SenseTime Open-Sources Unified Vision Model SenseNova-Vision
SenseTime has released SenseNova-Vision, a unified vision model designed to handle multiple computer vision tasks. The model integrates object detection, OCR, segmentation, and 3D reconstruction within a single system.
Tech company SenseTime has released and made open-source SenseNova-Vision, a unified vision model within its SenseNova foundation model suite. The model is engineered to process a range of computer vision tasks, including object detection, optical character recognition, keypoint localization, image segmentation, and depth estimation, all within a single system.
Historically, these functions would require separate, specialized models. SenseNova-Vision aims to consolidate these capabilities, potentially streamlining development and enhancing efficiency in AI applications. The model also supports dense geometric prediction, such as surface normal estimation, and multi-view 3D tasks like point cloud reconstruction and camera pose estimation.
SenseTime has also launched the SenseNova-Vision Corpus-50M, a visual instruction dataset comprising 50 million samples. The company intends to integrate the model into its SenseNova U-series in the future.
This release may influence the development trajectory of computer vision models and open-source contributions. The availability of open data and models could foster the creation of novel applications across various industries.