search

jaclyngu / FactorMatte

104 stars 7 forks source link

May I ask when the code will be uploaded, looking forward to trying it out #2

Open zhanghongyong123456 opened 1 year ago

jaclyngu commented 1 year ago

Thanks for your interest! I'll try to release it in the next month or two :).

franz101 commented 1 year ago

Amazing, let me know if you need help with creating the google colabs

jauri1 commented 1 year ago

I checked this repo every 3-4 days for 2 months now :(

jaclyngu commented 1 year ago

I checked this repo every 3-4 days for 2 months now :(

Sorry for the delay. Just pushed the first version.

shiron8bit commented 1 year ago

Hello, thank you for the code. As for custom dataset, do we still need to do point 3? I couldn't find any gen_pos_data.py script.

And if we need no to do it, what about logic in 'keypoint_homo_short.ipynb'? As i can see, it uses mask/01 folder (which we don't generate).

jaclyngu commented 1 year ago

Sorry for the confusion! I realized some files were missing. I just uploaded those and added more comments. Please refer to the updated README, and let me know if you have any questions! (Note: I might be a bit busy until the ICCV deadline.)

At 2023-02-27 23:08:46, "shiron8bit" @.***> wrote:

Hello, thank you for the code. As for custom dataset, do we still need to do point 3? I couldn't find any gen_pos_data.py script.

And if we need no to do it, what about logic in 'keypoint_homo_short.ipynb'? As i can see, it uses mask/01 folder (which we don't generate).

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

shiron8bit commented 1 year ago

Hello, even if the person is able to correct some of the scripts on their own (obsolete flags juggling, adding --get_bg for creating bg_gt.png after stage1, manually changing part of the code fow working with bg_gt in gen_backgroundPosEx instead of frames in sourcedir), it still hard to start stage2: errors in plausibleLoss - i guess it somehow connected to pos_ex_dirs creation, but i'm not sure.

Are you planning to provide a more generic and cleaner version of the code?