1604.Joint Detection and Identification Feature Learning for Person Search論文閱讀筆記

前言： Li et al. 和Ahmed et al. 提出方法：一對裁剪圖像輸入+binary verification loss 訓練參數 Ding et al. 和Cheng et al. 提出方法：三圖像輸入（triplet sample）+binary triplet loss 訓練參數，minimize the feature distance between the same per

>>阅读原文<<