search

dk-liang / FIDTM

[IEEE TMM] Focal Inverse Distance Transform Maps for Crowd Localization
MIT License
169 stars 41 forks source link

I-SSIM Loss #27

Closed BoltenWang-Meta closed 2 years ago

BoltenWang-Meta commented 2 years ago

Sorry, i didn' t find your implementation of I-SSIM loss.

feixuedudiao commented 2 years ago

the same too.

dk-liang commented 2 years ago

Sorry, after discussing with my coauthors, we can not release the I-SSIM as soon as possible before this work is formally published. Actually, the performance of the baseline can also be further improved by using some existing tricks.

rydenisbak commented 1 year ago

I did implement I-SSIM and add some useful tricks, I can publish it in my fork, if someone needs it. I got 55.86 MAE instead of 57.0 from paper on SHA dataset.

mariosconsta commented 1 year ago

I did implement I-SSIM and add some useful tricks, I can publish it in my fork, if someone needs it. I got 55.86 MAE instead of 57.0 from paper on SHA dataset.

If the authors allow it I would like to see the implementation of I-SSIM for personal usage.

rydenisbak commented 1 year ago

@mariosconsta hello Type of license of the code allows me be free to modify and publish code, so I will publish in couple days, because now i have some important deadline in my job

mariosconsta commented 1 year ago

@rydenisbak Hey man, I hope you are doing great. Just checking if you had the time to update your fork with the implementation.

Gananathan commented 1 year ago

@mariosconsta hello Type of license of the code allows me be free to modify and publish code, so I will publish in couple days, because now i have some important deadline in my job

Hey @rydenisbak, is there any update on publishing of the loss function module you have implemented. Thank you in advance.

Thepang25 commented 1 year ago

Hi,@rydenisbak I noticed that the author's paper has been successfully published, when can you update your reproduced I-SSIM loss. I have followed your Home page, and I really need your fantastic work! Thank you in advance.

rydenisbak commented 1 year ago

excuse me guys @mariosconsta @Gananathan @Thepang25 for looooong waiting,

I published my implementation, now you can check it out https://github.com/rydenisbak/FIDTM/tree/visionlabs_implement

rydenisbak commented 1 year ago

@dk-liang should I create PR ? May be you create new branch for this PR? Now I can do PR only to master branch