search

akshitac8 / BiAM

[ICCV 2021] Official Pytorch implementation for Discriminative Region-based Multi-Label Zero-Shot Learning SOTA results on NUS-WIDE and OpenImages
Apache License 2.0
60 stars 12 forks source link

About download_imgs. Py #6

Closed sorrowyn closed 3 years ago

sorrowyn commented 3 years ago

No such file or directory: 1:labelmap_path=path+'classes-trainable.txt' 2:dict_path=path+'class-descriptions.csv' 3:df_label = pd.read_csv(path+data_set+'/annotations-human.csv') 4:df_image = pd.read_csv(path+data_set+'/images.csv')

sorrowyn commented 3 years ago

https://github.com/hbdat/cvpr20_LESA/tree/master/data/OpenImages/2018_04 图片

sorrowyn commented 3 years ago

The download speed is too low to use proxy or not to use in china.

图片 图片

akshitac8 commented 3 years ago

Hello @sorrowyn, Its not possible for us to upload the openimages dataset on the drive. for this you need to contact the openimages dataset authors. OpenImages generally takes anywhere around 1 to 2 weeks to download because it has to fetch from Flickr.

sorrowyn commented 3 years ago

Hello @sorrowyn, Its not possible for us to upload the openimages dataset on the drive. for this you need to contact the openimages dataset authors. OpenImages generally takes anywhere around 1 to 2 weeks to download because it has to fetch from Flickr. Thank you @akshitac8 for your reply. Can I use this method to download the openimages dataset? https://github.com/cvdfoundation/open-images-dataset#download-images-with-bounding-boxes-annotations 图片

akshitac8 commented 3 years ago

Hello, @sorrowyn This will download data for the open-images detection dataset and not for classification. The detection dataset has training images of 1,743,042 while the classification has around 5M training files.

akshitac8 commented 3 years ago

@sorrowyn is the issue solved?

sorrowyn commented 3 years ago

@sorrowyn is the issue solved? Hello,@akshitac8. I am currently testing. This problem can be solved with your help.d

图片

sorrowyn commented 3 years ago

is 37600 images in validation? There are some errors: 1:<class 'PIL.UnidentifiedImageError'> 2:/home/XXX/anaconda3/envs/tensorflow/lib/python3.8/site-packages/PIL/Image.py:2911: DecompressionBombWarning: Image size (90642948 pixels) exceeds limit of 89478485 pixels, could be decompression bomb DOS attack.

pillow=8.3.1

图片

akshitac8 commented 3 years ago

Hello @sorrowyn The no. of images will depend on how many are left on Flickr for the openimages dataset it can different from what we used. For pillow please try with v7.2.0

sorrowyn commented 3 years ago

图片

akshitac8 commented 3 years ago

@sorrowyn is the issue solved?

sorrowyn commented 3 years ago

Hi, akshitac@akshitac8! Thanks to your sincere help, the problem was solved.

pillow = 8.3.1 n_jobs = 32 in download_imgs. Errors occur when n_job is greater than 32

akshitac8 commented 3 years ago

@sorrowyn happy to help :) I will be closing the issue now.

NanAlbert commented 2 years ago

Hello. @akshitac8 @sorrowyn I also meet the problem "<class 'PIL.UnidentifiedImageError'>" and "drop image", and it's also very slow. Can you provide detailed versions of any installed packages? Thanks a lot. Some potentially important packages: h5py 3.1.0 imageio 2.0.1 importlib-metadata 4.8.3 joblib 1.1.0 matplotlib 3.3.4 networkx 2.5.1 numpy 1.19.5 opencv-python 4.5.5.62 pandas 1.1.5 Pillow 8.3.1 scikit-image 0.15.0 scikit-learn 0.24.2 scipy 1.5.4 setuptools 59.6.0 requests 2.27.1 requests-oauthlib 1.3.1 image

Sen-Ran commented 1 year ago

@NanAlbert Have you solved this problem?

sorrowyn commented 1 year ago

@NanAlbert Have you solved this problem?

中间一部分数据无法下载，估计是连接失效了

Sen-Ran commented 1 year ago

@sorrowyn how many images have you already downloaded? and how much space did these images cost?

Sen-Ran commented 1 year ago

@NanAlbert Have you solved this problem?

中间一部分数据无法下载，估计是连接失效了

Did you download into the 2018_04 folder or into the 2017_11 folder?