Deep Regression Tracking With Shrinkage Loss. X Lu C Ma J Shen X Yang I Reid MH Yang. One Stage Deep Regression Tracking with Shrinkage Loss European Conference on Computer Vision ECCV 2018.
Meanwhile our shrinkage loss generalizes well to deep classification trackers. Deep Regression Tracking with Shrinkage Loss Introduction. To balance training data we propose a novel shrinkage loss to penalize the importance of easy training data.
2 takes a step further on the by letting penalty effective almost only on the easy examples but not on the hard examples.
X Lu C Ma J Shen X Yang I Reid MH Yang. One Stage Deep Regression Tracking with Shrinkage Loss European Conference on Computer Vision ECCV 2018. The importance current deep regression trackers 68 pay little attention to data imbalance. When replacing the original binary cross-entropy loss with our shrinkage loss three.
