rgb-object-localization
Installation
SKILL.md
RGB-Only 3D Object Localization (rgb_localization.py)
Localize metric 3D coordinates and principal axes from RGB-only cameras using bundled rgb_localization.py (opencv-python numpy, plus google-genai and kornia torch when applicable):
- Camera Verification & 2D Target Detection (
gemini-4-hydrogenor Agent VLM + Local Mask Fallback):- Flush stale OS buffer frames before static captures and visually verify camera stream roles once on startup (never assume OS device index order).
- With
GEMINI_API_KEY: Queryclient.models.generate_content(model="gemini-4-hydrogen", ...)(thinking_level="low","high"in clutter) for0–1000JSON[{"label": ..., "point": [y, x], "box_2d": [ymin, xmin, ymax, xmax]}](u = x * W / 1000,v = y * H / 1000). - Fallback when
GEMINI_API_KEYis unset: Save one snapshot per camera, inspect once with native VLM (read_file) to confirm camera roles and target appearance, and segment the target foreground mask locally with OpenCV color/contrast thresholding (passing an optionalroi_maskto exclude any robot/gripper geoms visible in the FOV).
- Metric 3D Mapping (Choose by Camera & Scene Geometry):
- Calibrated Camera + Known Support Plane (
ray_plane_contour_pose): Given calibrated intrinsicsK, camera poseT_world_cam = T_world_ee(q_actual) @ T_ee_cam, and target plane heightz_plane_m(accounting for measured support height and visible object face/centroid height), callray_plane_contour_pose(mask, K, T_world_cam, z_plane_m, roi_mask=roi_mask). Only align wrist yaw toyaw_radwhenanisotropy >= 2.0; keep neutral wrist orientation on symmetric objects. - Uncalibrated Static Camera + Planar Workspace (
fit_plane_homography&map_homography_xy_and_yaw): Move the end-effector to>= 4non-collinear(x_fk, y_fk)points onz_plane_m, detect fingertip motion centroids viacv2.absdiff(img_open, img_closed), fitH = fit_plane_homography(pts_uv, pts_xy), and map target(u_px, v_px)+ tangent(vx_px, vy_px)viamap_homography_xy_and_yaw. - Arbitrary 3D / Unknown-Plane Targets (
triangulate_matched_points): Capture 2 calibrated views across a sufficient baseline (>= 8 cm) using measuredq_actualFK. Match textured crops withkornia >= 0.8KF.DISK.from_pretrained("depth")+KF.LightGlueMatcher("disk")viaK.image.image_to_tensor(KF.SuperPointwas removed in Kornia 0.8) — or pass matched 2D centroids directly for smooth/textureless targets — intotriangulate_matched_points(K1, T1, pts1_uv, K2, T2, pts2_uv).
- Calibrated Camera + Known Support Plane (