圖像預處理中resize和padding
resize
如果只是爲了獲得要求尺寸的圖像,最常用的方法就是resize, the usual practice is to resize the input image to the given size(但是resize有缺點,the image aspect ratio is no longer kept.)
This practice may work well for image classification where fine details may not be necessary. But for image retrieval, we want to keep the image aspect ratio unchanged.
而行人重識別又是圖像檢索的子任務。依據是以下的知乎鏈接:
上面加粗的文字引用自如下博客: