Stitching live frames to 160 per second using C++
Project detail
– object contor class
– last frame list used in contor matching
– display frame as forward weighted average of detected frames
– self removal if not seen in X frames
– integration of opencv. object trajectory prediction by Y future frames
– area of incidence class
– verticle wedge to polygon and tangental intersection of prediction point
– CV frame class
– dynamic framedrop for behind frames
– test accuracy of monochromatic filter against detection engine
– test accuracy of quarter frame detection
– UI display
– refactor percentage of original vs resized frame vs XY coords
– refactor heat colours for criticality
– frame ingest engine
– polynominal stiching and stretching for clarity
– 3 source stitch