a bug

[guyii54]

this line should be cv::Point2f res = detect(_tmpl, getFeatures(image, 0, _scale_dsst), _peak_value);
otherwise the size of translation filter extracted_roi in getFeatures() is always the same(base_target_size*padding).

OpenTracker/kcf/kcftracker.cpp

Line 309 in d7fac97

cv::Point2f res = detect(_tmpl, getFeatures(image, 0, 1.0f), _peak_value);