visionaix / sam3-1
SAM 3.1 (Segment Anything 3.1): Unified promptable segmentation for images. Supports text, point, and box prompts. Detects and segments 270K+ concepts. By Meta FAIR.
visionaix / sam3
SAM 3.1 (Segment Anything 3.1): Unified promptable segmentation for images. Supports text, point, and box prompts. Detects and segments 270K+ concepts. By Meta FAIR.
visionaix / metric3dv2
Metric3D v2 (TPAMI 2024): Monocular metric depth and surface normals from a single image. Predicts real-world depth in meters. Works indoor and outdoor.
visionaix / metric3d-v2
Metric3D v2 (TPAMI 2024): Monocular metric depth and surface normals from a single image. Predicts real-world depth in meters. Works indoor and outdoor with ViT-Small and ViT-Large backbones.
visionaix / geocalib
GeoCalib (ECCV 2024): Single-image camera calibration. Estimates focal length, FoV, distortion, roll and pitch from one image using a deep net + Levenberg-Marquardt optimizer. Works on both outdoor and indoor scenes.