I hold on an opinion that if you need to do something for more than 10 times, just forget it and try to write codes. This is the basic idea of "tool of Li", it is a collection of small tools I write during my daily life. Parts of codes are borrowed from many website like github.com, csdn and even the search engine. If there is any problem, please contact me to remove them and thanks a lot for the people who share their codes. Of the people, for the people, feel free to use the codes as you like.
A small tool used to select images, when training neural network, sometimes we need to select the images which are good for our training data,here is the tool. You just need to cluster the img once and don't need to do it again,you can start and quit anytime. Press key 'c' to reject the img, press key 'q' to quit and any other key means you accept the img.
我一直认为如果一件事情,你需要做很多遍,那么你最好写个代码完成它,所以这就是我写了这个工具箱系列的原因。系列里面的代码基本都是我日常积累下来偷懒用的code。代码的来源很杂,基本都是我根据自己需求,结合github.com,CSDN,甚至是搜索得来的代码修改而来的(如果有任何问题,请联系我移除他们,其实代码中有很多原作者的信息之类,我一般都保留了)。取之于人,用之于人,所以这些代码大家自由使用,开心就好(当然,写的这么这么一般的代码估计都不会有人看,哈哈哈!!我英语太差,所以借着这个机会练练写英语,所以前面英语的部分,大家当没看到)。
这个工具是为机器学习或者深度学习生成数据集服务的,往往在前期需要对数据集进行删选,利用这个工具,输入任何键表示接受图片,输入c表示拒绝图片,输入q表示退出。对于筛选过的图片,程序会自动识别,不需要反复操作。所以可以在任何时候开始,任何时候结束,不用任何担心需要重复操作。代码从文件目录下读取所有后缀名为jpg和png的文件,并将它们放入accept和reject文件夹。