GLNet
GLNet copied to clipboard
Validation results
Hi, When I run your code I get the following validation mIoU accuracies with the pretrained models:
- mode 1 (global only): 63.9%
- mode 2: 69.2%
- mode 3: 69.9 %
Is this to be expected? I had to fix some issues when running the evaluation code (e.g. related to dataset split and https://github.com/TAMU-VITA/GLNet/issues/17 ) so there might be something wrong with my fixes. Which models/mode achieves the 71.6% accuracy from the paper? The global-only accuracy should also achieve 66.4% instead of 63.9%, right?
Hi @thomasverelst! Thank you very much for your interest in our work!
In our paper, we report the mIoU on the test set, not the validation set.
hi@thomasverelst! RuntimeError: The size of tensor a (48387) must match the size of tensor b (16129) at non-singleton dimension 0 when i put RGB images input to Network.can you tell how to train the model ?
@zzx0836 Hi! I changed some things in helper.py, but now it doesnt seem to run anymore (hangs on the first sample). If you want to take a look what I changed, I've pushed the changes to my fork https://github.com/thomasverelst/GLNet. Running the code was extremely slow anyway, maybe due to the fixes I added.
@zzx0836 Hi! I changed some things in helper.py, but now it doesnt seem to run anymore (hangs on the first sample). If you want to take a look what I changed, I've pushed the changes to my fork https://github.com/thomasverelst/GLNet. Running the code was extremely slow anyway, maybe due to the fixes I added.
the Deep Globe "Land Cover Classification" dataset , in this dataset crossvali, *mask_png is existing?
Can you tell me which functions you changed?
------------------ 原始邮件 ------------------ 发件人: "VITA-Group/GLNet" <[email protected]>; 发送时间: 2020年11月13日(星期五) 下午3:27 收件人: "VITA-Group/GLNet"<[email protected]>; 抄送: "1416055750"<[email protected]>;"Mention"<[email protected]>; 主题: Re: [VITA-Group/GLNet] Validation results (#27)
@zzx0836 Hi! I changed some things in helper.py, but now it doesnt seem to run anymore (hangs on the first sample). If you want to take a look what I changed, I've pushed the changes to my fork https://github.com/thomasverelst/GLNet. Running the code was extremely slow anyway, maybe due to the fixes I added.
the Deep Globe "Land Cover Classification" dataset , in this dataset crossvali, *mask_png is existing?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
Can you tell me which functions you changed? … ------------------ 原始邮件 ------------------ 发件人: "VITA-Group/GLNet" <[email protected]>; 发送时间: 2020年11月13日(星期五) 下午3:27 收件人: "VITA-Group/GLNet"<[email protected]>; 抄送: "1416055750"<[email protected]>;"Mention"<[email protected]>; 主题: Re: [VITA-Group/GLNet] Validation results (#27) @zzx0836 Hi! I changed some things in helper.py, but now it doesnt seem to run anymore (hangs on the first sample). If you want to take a look what I changed, I've pushed the changes to my fork https://github.com/thomasverelst/GLNet. Running the code was extremely slow anyway, maybe due to the fixes I added. the Deep Globe "Land Cover Classification" dataset , in this dataset crossvali, *mask_png is existing? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
can you give me the Deep Globe "Land Cover Classification" dataset ?thank you very much!
can you share the Deep Globe "Land Cover Classification" dataset with me? Research use,thank you
I didn't download this dataset. You could register and download the Deep Globe "Land Cover Classification" dataset here: https://competitions.codalab.org/competitions/18468
------------------ 原始邮件 ------------------ 发件人: "VITA-Group/GLNet" <[email protected]>; 发送时间: 2020年11月17日(星期二) 晚上7:11 收件人: "VITA-Group/GLNet"<[email protected]>; 抄送: "1416055750"<[email protected]>;"Mention"<[email protected]>; 主题: Re: [VITA-Group/GLNet] Validation results (#27)
can you share the Deep Globe "Land Cover Classification" dataset with me? Research use,thank you
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
I didn't download this dataset. You could register and download the Deep Globe "Land Cover Classification" dataset here: https://competitions.codalab.org/competitions/18468
------------------ 原始邮件 ------------------ 发件人: "VITA-Group/GLNet" <[email protected]>; 发送时间: 2020年11月13日(星期五) 晚上7:01 收件人: "VITA-Group/GLNet"<[email protected]>; 抄送: "1416055750"<[email protected]>;"Mention"<[email protected]>; 主题: Re: [VITA-Group/GLNet] Validation results (#27)
Can you tell me which functions you changed? … ------------------ 原始邮件 ------------------ 发件人: "VITA-Group/GLNet" <[email protected]>; 发送时间: 2020年11月13日(星期五) 下午3:27 收件人: "VITA-Group/GLNet"<[email protected]>; 抄送: "1416055750"<[email protected]>;"Mention"<[email protected]>; 主题: Re: [VITA-Group/GLNet] Validation results (#27) @zzx0836 Hi! I changed some things in helper.py, but now it doesnt seem to run anymore (hangs on the first sample). If you want to take a look what I changed, I've pushed the changes to my fork https://github.com/thomasverelst/GLNet. Running the code was extremely slow anyway, maybe due to the fixes I added. the Deep Globe "Land Cover Classification" dataset , in this dataset crossvali, *mask_png is existing? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
can you give me the Deep Globe "Land Cover Classification" dataset ?thank you very much!
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
hi@thomasverelst! RuntimeError: The size of tensor a (48387) must match the size of tensor b (16129) at non-singleton dimension 0 when i put RGB images input to Network.can you tell how to train the model ?
RuntimeError: The size of tensor a (96774) must match the size of tensor b (290322) at non-singleton dimension 0
How to fix it?