i modified a script but it seems to give this error
(voc_videostyle) C:\Users\Gebruiker\Documents\visionsofchaos\fewshot>python C:\\deepdream-test\\Few-Shot-Patch-Based-Training-master\\generate.py --checkpoint data\project2_train\logs_reference_P\model_00003.pth --data_root "data" --dir_input project2_gen\input_filtered --outdir=data\project2_gen\output --device="cuda:0"
device: cuda:0
Batch 0 / 116
Traceback (most recent call last):
File "C:\deepdream-test\Few-Shot-Patch-Based-Training-master\generate.py", line 54, in <module>
net_out = generator(net_in)
File "C:\Users\Gebruiker\AppData\Roaming\Visions of Chaos\Examples\MachineLearning\venv\voc_videostyle\lib\site-packages\torch\nn\modules\module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "C:\deepdream-test\Few-Shot-Patch-Based-Training-master\models.py", line 111, in forward
output_0 = self.conv0(x)
File "C:\Users\Gebruiker\AppData\Roaming\Visions of Chaos\Examples\MachineLearning\venv\voc_videostyle\lib\site-packages\torch\nn\modules\module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "C:\Users\Gebruiker\AppData\Roaming\Visions of Chaos\Examples\MachineLearning\venv\voc_videostyle\lib\site-packages\torch\nn\modules\container.py", line 141, in forward
input = module(input)
File "C:\Users\Gebruiker\AppData\Roaming\Visions of Chaos\Examples\MachineLearning\venv\voc_videostyle\lib\site-packages\torch\nn\modules\module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "C:\Users\Gebruiker\AppData\Roaming\Visions of Chaos\Examples\MachineLearning\venv\voc_videostyle\lib\site-packages\torch\nn\modules\conv.py", line 447, in forward
return self._conv_forward(input, self.weight, self.bias)
File "C:\Users\Gebruiker\AppData\Roaming\Visions of Chaos\Examples\MachineLearning\venv\voc_videostyle\lib\site-packages\torch\nn\modules\conv.py", line 443, in _conv_forward
return F.conv2d(input, weight, bias, self.stride,
RuntimeError: Given groups=1, weight of size [32, 6, 7, 7], expected input[1, 3, 720, 1280] to have 6 channels, but got 3 channels instead
this is the code for the .py file, i saw some other people have similar issues but mine is not being fixed by the answers they get.
does anyone know how to fix this or make it modular so i can use a args command for it if it changes with different generations?
import argparse
import os
from PIL import Image
from custom_transforms import *
import numpy as np
import torch.utils.data
import time
from data import DatasetFullImages
# Main to generate images
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--checkpoint", help="checkpoint location", required=True)
parser.add_argument("--data_root", help="data root", required=True)
parser.add_argument("--dir_input", help="dir input", required=True)
parser.add_argument("--dir_x1", help="dir extra 1", required=False)
parser.add_argument("--dir_x2", help="dir extra 2", required=False)
parser.add_argument("--dir_x3", help="dir extra 3", required=False)
parser.add_argument("--outdir", help="output directory", required=True)
parser.add_argument("--device", help="device", required=True)
args = parser.parse_args()
generator = (torch.load(args.checkpoint, map_location=lambda storage, loc: storage))
generator.eval()
if not os.path.exists(args.outdir):
os.mkdir(args.outdir)
device = args.device
print("device: " + device, flush=True)
generator = generator.to(device)
if device.lower() != "cpu":
generator = generator.type(torch.half)
transform = build_transform()
dataset = DatasetFullImages(args.data_root + "/" + args.dir_input, "ignore", "ignore", device,
dir_x1=args.data_root + "/" + args.dir_x1 if args.dir_x1 is not None else None,
dir_x2=args.data_root + "/" + args.dir_x2 if args.dir_x2 is not None else None,
dir_x3=args.data_root + "/" + args.dir_x3 if args.dir_x3 is not None else None,
dir_x4=None, dir_x5=None, dir_x6=None, dir_x7=None, dir_x8=None, dir_x9=None)
imloader = torch.utils.data.DataLoader(dataset, 1, shuffle=False, num_workers=1, drop_last=False) # num_workers=4
generate_start_time = time.time()
with torch.no_grad():
for i, batch in enumerate(imloader):
print('Batch %d / %d' % (i, len(imloader)))
net_in = batch['pre'].to(args.device)
if device.lower() != "cpu":
net_in = net_in.type(torch.half)
net_out = generator(net_in)
#image_space_in = to_image_space(batch['image'].cpu().data.numpy())
#image_space = to_image_space(net_out.cpu().data.numpy())
image_space = ((net_out.clamp(-1, 1) + 1) * 127.5).permute((0, 2, 3, 1))
image_space = image_space.cpu().data.numpy().astype(np.uint8)
for k in range(0, len(image_space)):
im = image_space[k] #image_space[k].transpose(1, 2, 0)
Image.fromarray(im).save(os.path.join(args.outdir, batch['file_name'][k]))
print(f"Generating took {(time.time() - generate_start_time)}", flush=True)
thanks