The largest and most realistic dataset available for TCC. It consists of 600 real-world videos recorded with a high-resolution mobile phone camera shooting 1824 x 1368 sized pictures. The length of these videos ranges from 3 to 17 frames (7.3 on average, the median is 7.0 and mode is 8.5). Ground truth information is present only for the last frame in each video (i.e., the shot frame), and was collected using a gray surface calibration target.
Paper | Code | Results | Date | Stars |
---|