search

Openning07 / MPADA

The official implementation of MP_{ada} in Attention-based Multi-patch Aggregation for Image Aesthetic Assessment (MM 2018)
79 stars 11 forks source link

"module 'dataset' has no attribute 'AVA2012"&"test in another dataset" #10

Open cherryolg opened 2 years ago

cherryolg commented 2 years ago

Hi, I have a few questions

  1. "You need to put the ava2012. Py in AVA_info in the folder of tensorpack. dataflow. dataset." how to operate it? Can you explain it in detail? 2.AttributeError: module 'dataset' has no attribute 'AVA2012. This error is reported when I run, may be 1 without processing?
  2. I want to test my own data set and get an aesthetic score. What do I need to change? Looking forward to your answer, thank you!
Openning07 commented 2 years ago

Thanks for your comments.

  1. Since the project is built on tensorpack lib, and we need to create a new data loader to load the AVA benchmark. The file of ava2012.py is https://github.com/Openning07/MPADA/blob/master/AVA_info/ava2012.py. The tensorpack.dataflow.dataset is tensorpack/dataflow/dataset (the directory structure of tensorpack lib).
  2. To test on your own data, you can i) prepare it in the same format as AVA dataset and use ava2012.py to load the data; or ii) write another data loader py file, which is similar to ava2012.py. To get the aesthetic score, please refer to https://github.com/Openning07/MPADA/issues/8.

Hope that the information helps you solve your problem.

cherryolg commented 2 years ago

Thanks a lot for your reply.  Forgive my stupidity, I still have some questions to ask you.    1. I want to input my own dataset and get aesthetic scores like in MPADA-master\AVA_info\AVA_scores.txt.  Do I need to add "--eval"？  "You can get the visual aesthetic scores by fetching the classification confidences of the positive class." I'm sorry I still don't understand the meaning of this sentence, can you explain it more specifically?    2. When I run "--eval", a "val.txt" file will be used. Is this the "AVA_scores.txt" in 1?  When I rename AVA_scores.txt to val.txt, it works, but it reports " File "D:\anacode\envs\tensorflow18\lib\site-packages\tensorpack\dataflow\parallel.py", line 109, in init assert  os.name != 'nt', "ZMQ IPC doesn't support windows!" AssertionError: ZMQ IPC doesn't support windows!" Does the program have to run on Linux?    3. Should I add "--eval" to get an aesthetic score?  I added "--eval" and AVA_test.lst was not used when running.    Hope to hear from you again!

------------------ 原始邮件 ------------------ 发件人: "Openning07/MPADA" @.>; 发送时间: 2022年4月12日(星期二) 晚上7:07 @.>; @.**@.>; 主题: Re: [Openning07/MPADA] "module 'dataset' has no attribute 'AVA2012"&"test in another dataset" (Issue #10)

Thanks for your comments.

Since the project is built on tensorpack lib, and we need to create a new data loader to load the AVA benchmark. The file of ava2012.py is https://github.com/Openning07/MPADA/blob/master/AVA_info/ava2012.py. The tensorpack.dataflow.dataset is tensorpack/dataflow/dataset (the directory structure of tensorpack lib).

To test on your own data, you can i) prepare it in the same format as AVA dataset and use ava2012.py to load the data; or ii) write another data loader py file, which is similar to ava2012.py. To get the aesthetic score, please refer to #8.

Hope that the information helps you solve your problem.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>