sounakdey / doodle2search

Doodle to Search: Practical Zero Shot Sketch Based Image Retrieval
https://sounakdey.github.io/doodle2search.github.io/
74 stars 25 forks source link

can't reproduce accurate MAP #9

Open buptlmk opened 4 years ago

buptlmk commented 4 years ago

Respected Authors, Your paper is great. I have two question to ask you.
The first one is that after training on Sketchy-Extend dataset, I can't reproduce your result. Your mAP result is 0.3691 in the paper, but I only got 0.0589. Can you publish your pretrained model? The second one is I can't produce sketchy_semantic_label by word2Vec, it told me "word 'XXX' is not in vocabulary"。Can you help me deal with this problem? Thank you!

haowang1992 commented 4 years ago

(1) I got 0.058 map reproduced, too. (2) I think the word xxx should be seperated, for example, car_(sedan)

payne53 commented 4 years ago

I got 0.363 map, but I use the semantic_label released by the author.

haowang1992 commented 4 years ago

@payne53 Thank you so much for pointing it. Could you reproduce the performance on QuickDraw dataset?

payne53 commented 4 years ago

@payne53 Thank you so much for pointing it. Could you reproduce the performance on QuickDraw dataset?

Sorry, I haven't tried QuickDraw yet.

dlai6620 commented 4 years ago

how did you guys get this running? I kept getting TypeError: 'int' object is not iterable errors after Preparing Data.

develop-productivity commented 2 years ago

I got 0.363 map, but I use the semantic_label released by the author.

Hi. Have you made any changes to the code?I just got 0.3294 map.I have not made changes to this code

buptlmk commented 2 years ago

您好~我已经收到您的邮件,我会尽快查看,祝生活愉快~