liuziwei7 / fashion-landmarks Goto Github PK

View Code? Open in Web Editor NEW

350.0 15.0 98.0 4.92 MB

Fashion Landmark Detection in the Wild

Home Page: http://personal.ie.cuhk.edu.hk/~lz013/projects/FashionLandmarks.html

MATLAB 100.00%

computer-vision deep-learning landmark-detection vision-for-fashion

fashion-landmarks's People

Contributors

Stargazers

Watchers

Forkers

jdc08161063 caomw tendence benjamesbabala hassyma cheneyfan lyk125 adrianhust kuanhoong booyang-cs codeashu yangsuo coocoky silasxue salopge c1a1o1 twinsyssy1018 6676401088 jrcondenast jurocix lavanyashukla chenbangfeng wyw636 yiqinggit walkoncross thiagoneves sedaym thinkronize kif2006 hejiaget labimage joke-dk walter1218 csisheng leiyuanma brianzhu01 chenhanxmu luolongqiang 110awesome haokaixuan xiaodongww harryzj koulihong311 vax521 zhangxujinsh wuguowuge last-one siyubian xflee hao0313 ssssssssssss reloadbrain mbencherif ml-lab lilin19890401 shatha2014 zhijl loverain1997 jerrylee80 aiaini66 wangerniuniu soccergame dnaport22 zhuguangqiang q5390498 slaysd anhvaut justin0242 sophieguan zwcdp ys7yoo kushalprak sudohello jxncyym chaoso denischiciudean anglebinbin yuzhangiot glorytune pkusnail ank-22 youngstu wuxiaochuang ella77 joyce725 melodylail man1029 ahmad4633 oftenliu jaivinjames isazheng888 peternara arkasarkar19 heiheisoftware genhao3 sky9743 sohaib0399 tieck-it

fashion-landmarks's Issues

论文里面的类中心到底是啥

你能再解释下吗：
f(k) -> R^(k x 1) where k=1...K
是说f(1)->R^(1x1)？我想不是吧。。。
弄懂这个很重要你能用通俗的语言描述下吗？什么是k，K，C，谢了！
还有f，i，C的维度是怎样的？

Getting landmark coordinates

Dear Ziwei,

Awesome and interesting work!

Do you mind shed some lights on what is purpose of the statement -get_orig_coordinate = @(p)((p+0.5)*224-repmat([offset(2),offset(1)]',[pipline.num_points,1]))/scale;?

I can not relate this to the paper, particularly (p+0.5)*224. I don't have matlab so I won't be able to debug the value but when I run pyCaffe those values of landmarks came out of stage 1 forward are very small, same as pseudo labels (all lower than 0.01). Any explanations will be greatly appreciated! Thanks.

Model about Consumer to Shop

Dear Ziwei Liu:
I have read your paper:DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations， and it is very helpful for my research.
You mentioned four benchmarks in your paper but I only found "fashion-landmarks" and "landmark detection" models in your GitHub.
Could you release the model about "Consumer to Shop" ?
It will very helpful for my research.
Thank you very much.

Run demo.m error

error pipline_show_results (line 15)
    img_show(224-x+1:224*2-x,224-y:224*2-y,:) = img;

error demo (line 35)
    pipline_show_results(img, prediction);

Unrecognized function or variable 'caffe_'.

I got the error while I run demo.m
How can I solve this?

Can we load the given models on cpu-machine for inference on new images

Dear Ziwei Liu
I am trying to run inference of the models available on github on some images. Please excuse if it's a naive question but can these models be loaded on cpu-machine to run inference ?

why the code link of DLAN is invalid now

https://github.com/yysijie/DLAN/

Loss functions

Where are the lossfunctions of the net? I need help in setting up that weighted L2 loss for landmark localisation

CelebA dataset query

Dear liu,
when I use the famous dataset CelebA, I found a problem: the location in list_bbox_celeba.txt is wrong, it does not fit the face. hope you will see my issue and tell me why this happened. waiting for you.

https://pan.baidu.com/s/1eSNpdRG#list/path=%2F

Best wishes

Can we use the bounding box in the testing set?

Dear Ziwei Liu:
A problem confuses me.Can we use the bounding box position as a konwn information in the testing set. I notice that you trian three network for upper,lower and full-body clothes. As I can not find the train code. Do you use the bounding box information to pre-process the training data.(crop the bounding box and then padding or resizing to 224*224?).If so,should we train a network to catch the bounding box position?(each class having an object detection network for the bounding box coordinate so that we can test a image and crop the information from the bounding box?).But the object detection network seems difficult to train because we can find that some images both having the upper clothing information and the lower clothing information, how can the network resolve what is need to recognition?Thanks for your answer.

fc8 or fc8_softlabel?

Dear @liuziwei7 : I am currently converting the demo.m from matlab to python since I don't have matlab. I am currently using the output from 'fc8' layer after forwarding the image. But output from fc8 are in negative float values. Can you confirm if this is as expected.

Model files are temporarily unavailable

Hi the model files are unavailable. Kindly inform when can they be available. Thanx