Open brandonhackney opened 7 months ago
The position data is temporally misscaled relative to the video data - that is, the position data first moves AFTER the video, but by the end moves BEFORE.
After intensive review, we've determined that only half of the position data the original authors provided is valid. Most of the leftover can be made valid with some manual adjustment, but 10% is just flat-out unusable. So we're giving up on the provided data. However, an ROI analysis is still a critical part of this experiment, so we're turning to computer vision applications to extract the position data for us instead.
TriCOPA comes with XY position data for each element on screen. Sajjad modified some of the videos for timing etc, so we need to verify that that data is still accurate.