search

waldo-seg / waldo

image-segmentation and text-localization
Apache License 2.0
13 stars 13 forks source link

[WIP] Add post-processing code (segmenter) #7

Closed hhadian closed 6 years ago

danpovey commented 6 years ago

@hhadian, I'm asking Yiwen to address any issues in the code that we've talked about here and make his own PR on top of it-- it's 2am in Iran so I guess you're sleeping now. This will help get Yiwen familiar with this part of the code.

hhadian commented 6 years ago

@hhadian, I'm asking Yiwen to address any issues in the code that we've talked about here and make his own PR on top of it-- it's 2am in Iran so I guess you're sleeping now. This will help get Yiwen familiar with this part of the code.

Yiwen did some code restructuring and cleaning in a separate PR (to my repo) and I fixed the issues in this PR myself because I was excited to see how it works. I guess we can merge this PR so Yiwen and I can separately work on it and to make things go faster.

@adelra, this is where we told you before that we need visualization. We'll probably be merging this today. You can pull the latest changes in Waldo and run segment.py (run training code first, i.e. run.sh). I guess ideally we need the algorithm's visualize() function to generate a .gif file which shows the progress of the algorithm from the beginning to the end (e.g., one image frame after every 1000 iterations).