Image recognition and camera positioning with OpenCV. A tourist guide application.
summary
OpenCV Python bindings perform SIFT/SURF-style keypoint detection and descriptor matching, with a SURF approximation used for speed; camera calibration uses chessboard photos to get intrinsics and undistort images; a homography from matched keypoints locates the picture corners and a solvePnP and Rodrigues step yields rotation/translation to reject false positives. Processing runs server-side for a Google Glass tourist guide, tested on about a 10-image library in the Palatino site; images must be in focus and multiple library views are required for robust 3D cases.
AI-written summary · about
chapters (6)
- 0:00Start
- 0:49Image recognition issues
- 5:46Extract keypoints and descriptors
- 6:15Same process for library image
- 6:27Match descriptors
- 13:10Image recognition is used for
related talks
22mHI and AI
21mLeveraging Python for Real-Time Image processing
40mThe eye of the Python, an eye tracking system. From zero to...
27mFrom minutes to seconds: Capillary auto-alignment with python and opencv
01h 33mWorkshop: Setting up Computer Vision with Kivy using Tensorflow
06mMaze Solver using Python and OpenCV
38mFinding Lane Lines for Self Driving Cars
20mScalable OCR pipelines using Python,Tensorflow & Tesseract