mi150 / RegenHance

4 stars 0 forks source link

Some problems when reproducing the code #2

Open husterwang opened 1 week ago

husterwang commented 1 week ago

Hi, an awesome work! I am trying to reproduce the "macroblock-based region importance predictor" and there are some problems in the code.

In "train_AccSR.py", 1、 the variable "sr_images_direc" and "o_images_direc" seem not to be used anywhere. 2、And in line 424, the path Test(root='/data01/wangweijun/accsr/AccMPEG/') is required to be revised. 3、In line 81, 88, 89, the data["read_id"], data["sr"], data["in"] will occur an error, since there is not such attribute in the "data" dictionary type according to the definition of class TEST.

Hope you can give me a hand. Thanks!

@censhallwe

censhallwe commented 1 week ago

Thanks, I see. I am sorry for the inconvenience caused by my failure to maintain the code for too long. I will now organize it and reply to you before 4pm China Standard Time on September 28th.

感谢您的回复,抱歉由于我太久没维护代码,给您造成了这些困扰。我现在整理一下,在北京时间9月28日(明天)下午4点前回复您!❤ @husterwang

censhallwe commented 1 week ago

Hello, thank you for your correction. I have reorganized the README document in the train directory and the code in the train_AccModel directory, and reproduced the process of steps 1-3. You can try again. If you encounter any problems, please feel free to leave a message below and @ me. I will provide an answer within 24 hours.

你好,感谢您的指正,我已经重新整理了train目录中的README文档以及train_AccModel目录下的代码,并复现了其中1-3步的流程,您可以再次进行尝试,如果遇到问题,欢迎您在下方留言并@我,我会在24小时内给予解答。 @husterwang

husterwang commented 1 week ago

Thank you very much for your help. I have no problem testing the training model, but there are still some problems with mask generation. In the (line 65-113) "train_AccModel/compress_blackgen_roi.py" file, the mask value seem not to be changed since it is initialized, and all of the values of it is saved as 1. Could you please look into this issue?@censhallwe

censhallwe commented 1 week ago

Sorry, it's my fault. You are right. 😵 I only tested the generation process and forgot to check the generated content. At present, I have revised and uploaded the file 'compress_blackgen_roi.py' again. I have deleted 'mask_slice' in line 66 and replaced 'mask_slice' in line 100 with 'mask'. You can download it from the code repository and replace the original file. Sorry for the inconvenience caused to you.🙇 If you encounter any other problems, please feel free to contact me at any time!🥳

抱歉,我的问题,您说得对。😵我只测试了生成的过程,忘记检查生成的内容了,目前我已重新修改并上传了compress_blackgen_roi.py文件,删去了第66行的mask_slice并把第100行的mask_slice换成了mask。您可以从代码库里下载下来后替换原本的文件,抱歉给您带来了困扰。🙇 如果遇到其他问题,欢迎随时联系我!🥳 @husterwang