Expert-level computer vision for robotics covering camera models, feature detection, object detection, pose estimation, visual SLAM, and deep learning for perception.
中文简介正在准备中。当前显示原文说明。
实测评分 未实测
查看安装方法安装前可以查看交给 AI 代理的指令内容。
Pinhole camera: projects 3D point to 2D image plane through focal point. Intrinsic matrix K: focal lengths and principal point, 5 parameters. Distortion: radial and tangential lens distortion, correct before processing. Stereo camera: two calibrated cameras, disparity gives depth. Depth cameras: structured light or time-of-flight, direct depth measurement.
Corners: Harris detector, stable under rotation, used for tracking. SIFT: scale and rotation invariant features, robust but slow. ORB: binary features, fast, used in real-time SLAM. Feature matching: descriptor distance, ratio test, RANSAC for outlier rejection.
Classical: HOG plus SVM, sliding window, deformable parts model. YOLO: single-shot detector, real-time, divides image into grid. Faster R-CNN: region proposal network, accurate but slower. Segmentation: semantic labels per pixel, instance segmentation separates objects.
SLAM: simultaneous localization and mapping, build map while tracking position. Visual odometry: estimate motion from consecutive image frames. Loop closure: detect revisited places, correct accumulated drift. ORB-SLAM3: feature-based, monocular and RGB-D, open-source reference system. NERF and 3DGS: neural radiance fields for photorealistic scene reconstruction.
| Pitfall | Fix |
|---|---|
| Skipping camera calibration | Always calibrate, reprojection error should be below 1 pixel |
| Ignoring lighting variation | Test and train under diverse lighting conditions |
| Assuming static scene in SLAM | Handle dynamic objects explicitly |
| Overconfident pose estimates | Always provide uncertainty with pose estimates |
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Expert-level abstract algebra knowledge. Use when working with groups, rings, fields, modules, Galois theory, representation theory, or algebraic structures. Also use when the user mentions 'group', 'ring', 'field', 'homomorphism', 'isomorphism', 'normal subgroup', 'quotient group', 'Sylow theorem', 'Galois theory', 'polynomial ring', 'ideal', 'module', 'vector space', or 'representation theory'.
中文简介正在准备中。当前显示原文说明。
Expert-level acoustics knowledge. Use when working with sound waves, acoustic propagation, resonance, standing waves, Doppler effect, room acoustics, ultrasound, noise control, musical acoustics, or architectural acoustics. Also use when the user mentions 'sound wave', 'decibel', 'frequency', 'resonance', 'standing wave', 'Doppler effect', 'acoustic impedance', 'reverberation', 'ultrasound', 'noise control', 'harmonics', or 'sound intensity'.
中文简介正在准备中。当前显示原文说明。
Expert-level aerodynamics covering subsonic and supersonic flow, lift and drag, airfoil theory, boundary layers, compressible flow, and CFD methods.
中文简介正在准备中。当前显示原文说明。
Expert-level air quality engineering covering pollutant sources, atmospheric dispersion, air quality standards, emission controls, and air quality modeling.
中文简介正在准备中。当前显示原文说明。
Expert-level aircraft structures covering structural design philosophy, wing and fuselage analysis, fatigue and damage tolerance, composite structures, and certification.
中文简介正在准备中。当前显示原文说明。
Expert-level algorithms and complexity theory covering sorting, searching, graph algorithms, dynamic programming, greedy algorithms, NP-completeness, and algorithm design paradigms.
中文简介正在准备中。当前显示原文说明。