Zhedong Zheng
Zhedong Zheng
Hi @zgplvyou ff1 is query feature ff2 is gallery feature We evaluate the result by the matching score. So we use sort
In market1501, I followed the original code using (x-y)^2 In cuhk, I rewrite the code using x*y (x,y have been normalized) since x^2 =1, y^2=1
The original setting is single shot. Every time we select 1 query and 100 images from 100 person (another camera . For every person, we randomly selects one image) as...
Hi @daichenchao You need to generate it. I wrote a prepare script. You may check here.https://github.com/layumi/2016_person_re-ID/blob/master/cuhk03-prepare-eval/prepare_imdb.m#L9
Hi @ycwychen Sorry for the late response. 1. Oxford5k is the test set. You can download it from http://www.robots.ox.ac.uk/~vgg/data/oxbuildings/ 2. I use the training set in **CNN Image Retrieval Learns...
Hi @hsqaihkl Did you change the training code for dukemtmc-reid? Or you just used the model trained on Market-1501?
No worries @hsqaihkl You may check your code with my baseline code. https://github.com/layumi/DukeMTMC-reID_baseline
Hi @xgh45 , Sorry for late response. What do you mean 'the query and test images are different'? Have you extracted the features? Please check the feature (all zeros? or...
Hi @sde123, I added some layers to matconvnet and I also included these layers in this repos. In fact, you do not need to install the original Matconvnet. I have...
I haven't met such error. Would you like to provide the whole log?