Online Classifier
ChenYang-1996 opened this issue · comments
A great job! And could you please say when the complete code(online Classifier included) will be uploaded? Thank you.
hello, in the vot official website, there exists the complete code including online classification.
相对于ATOM算法，我们调整了在线分类器的卷积核大小，将原来的4x4调整为了7x7，并且没用padding这这样就能够使得在线分类器输出的得分图和siamese输出的响应图大小一致，或者可以修改成两个4x4卷积核，为了测试方便可以把在线分类器随机初始化修改成固定初始化方式，可以让测试的结果完全固定下来@YangChen-UCAS 集成方式可以参考DROL的代码，或者也可以自己把ATOM的代码剥出来，直接塞进去，这样子代码就会很乱，不过我们就是这个做的，所以一直没有放出来
different from ATOM, we change the Convolution kernel of online classification from 4x4 to 7x7, and in order to get the same spatial size as obtained from siamese tracker, we remove the padding of kernel. Also, you can use two 4x4 convolution kernel. You can integrate the online classification according to the code of DROL, also you can extract the online classification code from ATOM directly, we use the second way, but it will make the code in a mess
Thank you for your reply, best wishes.