search

destiny301 / dpr

[WACV2024] Patch-based Selection and Refinement for Early Object Detection
17 stars 3 forks source link

How to obtain the complete code? #3

Open Lucy-Dina opened 7 months ago

Lucy-Dina commented 7 months ago

Thank you very much for the code of model training and patch selection you have shared. However, there is no Patch-Refiner module and Patch-Organizer module in the current code. As a beginner, I am really looking forward to having a complete code as a reference to help me understand algorithm principles and workflow. Meanwhile, the complete code enables me to visually evaluate the performance of the current algorithm on my own dataset based on the final image output, as shown in the flowchart. Therefore, will the complete code including Patch-Refiner module and Patch-Organizer module be added later? This will be very meaningful to me and I am looking forward to your reply and help. Thank you very much again.

destiny301 commented 7 months ago

Hi @Lucy-Dina

Thanks for your interest. Regarding your issue, I will update this repo later. But it may take some time for me to clean the codes. For the patch-refiner, you can first check this repo: https://github.com/Janspiry/Image-Super-Resolution-via-Iterative-Refinement This repo provides detailed instructions.

If you need further support, please let me know.

Lucy-Dina commented 7 months ago

Thank you very much for your reply and willingness to help me. I am looking forward to your update. Thank you again!

Lucy-Dina commented 6 months ago

Hello, based on the reference repo method you recommended earlier, I have roughly completed the patch-refiner module. But when I tried to connect it with the previous patch-selector module to form a complete DPR system, my output results were very poor. I'm not sure if there's a problem with my code logic, or if I missed some steps, or if there's a problem with my parameter settings. Therefore, I sincerely hope that you can provide the complete code of the whole DPR system for reference at your convenience, which will be very helpful in solving my current problem. I am looking forward to your reply and help. Thank you very much again.

destiny301 commented 6 months ago

Hi @Lucy-Dina ,

Sorry to hear your bad experience, and sorry for the late reply. Have you tried to finetune the patch-refiner model with your own dataset? If not, please try it. The results will be much improved after the finetuning.