Mike Azatov

Sorry not. For me, I was purposely clustering on more clusters to get the small clusters working.So before self-labeling step I needed to manually merge classes. Never figured out how...

As I understand larget `std` should result in larger `covariance` and hence smaller `gating_distance` but they don't seem to have much of an impact.

Have you had any luck doing that? @kennethchanchifung I'm working on a similar problem.

@muxizju Just came across this as I also have few classes. My question is what happens with the rest of the classes if I say I have 7 classes and...

@lamhoangtung , Were you able to figure this out? I'm trying to follow your steps to generate embeddings and compare them. But so far I'm running into some errors: I...

@lamhoangtung I think I figured out the first problem. 1. cv2 loads the image in BGR style so you need to convert it to RGB. 2. There seem to be...

I just came across this discussion. What's the significance of 0.693 number? @HuaZheLei @ergysr I was trying to train on Market1501 but with `fc1024_normalize`. The model converged for `fc1024` but...

Thanks @Pandoro for a prompt reply! 👍 You are correct, I didn't do anything special to load a pretrained network. I was wondering, did you mean to load the pretrained...

Thanks, I don't know how I missed that 🤦‍♂ 

I couldn't find your reported MAP scores anywhere. Does this look normal to you for Market dataset? ``` mAP: 65.72% | top-1: 81.71% top-2: 87.53% | top-5: 93.17% | top-10:...