shape issues

[ZubairKhan001]

File "/opt/OpenGait/opengait/data/transform.py", line 71, in call
return (x - self.mean) / self.std
ValueError: operands could not be broadcast together with shapes (30,64,64) (1,3,1,1)

[chuanfushen]

It looks like you want to apply BaseRgbTransform on your input. Then make sure your input is RGB images, which contains 3 channels for each pixel.

[ZubairKhan001]

Thank you, actually I input the pickle files folder.

…

On Wed, Jan 8, 2025 at 12:40 PM chuanfushen ***@***.***> wrote: It looks like you want to apply BaseRgbTransform on your input. Then make sure your input is RGB images, which contains 3 channels for each pixel. — Reply to this email directly, view it on GitHub <#260 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHEDVBHBPPKVAXDKTLQOKND2JTI5JAVCNFSM6AAAAABUZET27KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNZWHE2DIOJUGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[ZubairKhan001]

when i input RGB images folder File "/opt/OpenGait/opengait/data/dataset.py", line 43, in __loader__ raise ValueError('- Loader - just support .pkl !!!') ValueError: - Loader - just support .pkl !!! On Wed, Jan 8, 2025 at 3:30 PM Khan Engr zubair ***@***.***> wrote:

…

Thank you, actually I input the pickle files folder. On Wed, Jan 8, 2025 at 12:40 PM chuanfushen ***@***.***> wrote: > It looks like you want to apply BaseRgbTransform on your input. Then make > sure your input is RGB images, which contains 3 channels for each pixel. > > — > Reply to this email directly, view it on GitHub > <#260 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AHEDVBHBPPKVAXDKTLQOKND2JTI5JAVCNFSM6AAAAABUZET27KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNZWHE2DIOJUGI> > . > You are receiving this because you authored the thread.Message ID: > ***@***.***> >

[ZubairKhan001]

dataset_root: which path should be provided .pkl files/RGB . On Wed, Jan 8, 2025 at 4:07 PM Khan Engr zubair ***@***.***> wrote:

…

when i input RGB images folder File "/opt/OpenGait/opengait/data/dataset.py", line 43, in __loader__ raise ValueError('- Loader - just support .pkl !!!') ValueError: - Loader - just support .pkl !!! On Wed, Jan 8, 2025 at 3:30 PM Khan Engr zubair < ***@***.***> wrote: > Thank you, actually I input the pickle files folder. > > > On Wed, Jan 8, 2025 at 12:40 PM chuanfushen ***@***.***> > wrote: > >> It looks like you want to apply BaseRgbTransform on your input. Then >> make sure your input is RGB images, which contains 3 channels for each >> pixel. >> >> — >> Reply to this email directly, view it on GitHub >> <#260 (comment)>, >> or unsubscribe >> <https://github.com/notifications/unsubscribe-auth/AHEDVBHBPPKVAXDKTLQOKND2JTI5JAVCNFSM6AAAAABUZET27KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNZWHE2DIOJUGI> >> . >> You are receiving this because you authored the thread.Message ID: >> ***@***.***> >> >

[chuanfushen]

you can follow the structure of SUSTech1K dataset, putting all of your sequences within separated pkl.
https://github.com/ShiqiYu/OpenGait/tree/master/datasets/SUSTech1K

Then be noticed to modify the corresponding data_in_use as:

OpenGait/configs/lidargait/lidargait_sustech1k.yaml

Line 6 in 086f9f0

data_in_use: [false, true, false, false, false, false, false, false, false, false, false, false, false, false, false, false]

[ZubairKhan001]

ok thankyou, and what about the transforms, i used the data_in_use: [fasle,true], and a single transform instead of 2 transforms, getting this error File "/opt/OpenGait/opengait/modeling/models/BigGait.py", line 215, in forward ratios = ipts[1] # real_image_ratios shape: [n,s,ratio]; ratio: w/h, e.g. 112/224=0.5; IndexError: list index out of range

…

On Thu, Jan 9, 2025 at 10:41 AM chuanfushen ***@***.***> wrote: you can follow the structure of SUSTech1K dataset, putting all of your sequences within separated pkl. https://github.com/ShiqiYu/OpenGait/tree/master/datasets/SUSTech1K Then be noticed to modify the corresponding data_in_use as: https://github.com/ShiqiYu/OpenGait/blob/086f9f0129aa07c7b31df73593f07326c0ada26d/configs/lidargait/lidargait_sustech1k.yaml#L6 — Reply to this email directly, view it on GitHub <#260 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHEDVBEQMTM2LSQGS3RDPMD2JYDXJAVCNFSM6AAAAABUZET27KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNZZGIZDCOJZG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[chuanfushen]

BigGait requires two inputs, one is rgb modality, another is ratios.

OpenGait/opengait/modeling/models/BigGait.py

Lines 214 to 215 in 086f9f0

	sils = ipts[0] # input_images; shape: [n,s,c,h,w];
	ratios = ipts[1] # real_image_ratios shape: [n,s,ratio]; ratio: w/h, e.g. 112/224=0.5;

Follow BigGait configuration.yaml

OpenGait/configs/biggait/BigGait_CCPG.yaml

Line 6 in 086f9f0

data_in_use: [True, True] # images / real_ratios

Try to include ratios within input as well.

[zjq1839]

BigGait requires two inputs, one is rgb modality, another is ratios.

OpenGait/opengait/modeling/models/BigGait.py

Lines 214 to 215 in 086f9f0

sils = ipts[0] # input_images; shape: [n,s,c,h,w];
ratios = ipts[1] # real_image_ratios shape: [n,s,ratio]; ratio: w/h, e.g. 112/224=0.5;
Follow BigGait configuration.yaml

OpenGait/configs/biggait/BigGait_CCPG.yaml

Line 6 in 086f9f0

data_in_use: [True, True] # images / real_ratios

Try to include ratios within input as well.

Hello there, when I tried to try to organize my datasets according to the CCPG format, I found that the model actually loaded the 'subject-seqtype-view-aligned-sils.pkl', which is not the rgb one, is this correct?

I forced the model to load rgb images by setting 'data_in_use:[fasle,true,true,false], sils = ipts[1], ratios = ipts[0]', then I encounterd with a new error
[rank0]: File "OpenGait/opengait/utils/common.py", line 106, in ts2var
[rank0]: return autograd.Variable(x, **kwargs).cuda()
[rank0]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]: RuntimeError: Only Tensors of floating point and complex dtype can require gradients