levindabhi / cloth-segmentation Goto Github PK

This repo contains code and a pre-trained model for clothes segmentation.

License: MIT License

Python 100.00%

image-segmentation deep-learning cloth-segmentation image-parsing fashion pytorch clothing-parsing u2net

cloth-segmentation's Introduction

Clothes Segmentation using U2NET

This repo contains training code, inference code and pre-trained model for Cloths Parsing from human portrait.
Here clothes are parsed into 3 category: Upper body(red), Lower body(green) and Full body(yellow)

This model works well with any background and almost all poses. For more samples visit samples.md

Techinal details

U2NET : This project uses an amazing U2NET as a deep learning model. Instead of having 1 channel output from u2net for typical salient object detection task it outputs 4 channels each respresting upper body cloth, lower body cloth, fully body cloth and background. Only categorical cross-entropy loss is used for a given version of the checkpoint.
Dataset : U2net is trained on 45k images iMaterialist (Fashion) 2019 at FGVC6 dataset. To reduce complexity, I have clubbed the original 42 categories from dataset labels into 3 categories (upper body, lower body and full body). All images are resized into square ¯\_(ツ)_/¯ 768 x 768 px for training. (This experiment was conducted with 768 px but around 384 px will work fine too if one is retraining on another dataset).

Training

For training this project requires,

PyTorch > 1.3.0
tensorboardX
gdown

Download dataset from this link, extract all items.
Set path of train folder which contains training images and train.csv which is label csv file in options/base_options.py
To port original u2net of all layer except last layer please run python setup_model_weights.py and it will generate weights after model surgey in prev_checkpoints folder.
You can explore various options in options/base_options.py like checkpoint saving folder, logs folder etc.
For single gpu set distributed = False in options/base_options.py, for multi gpu set it to True.
For single gpu run python train.py
For multi gpu run
python -m torch.distributed.launch --nnodes=1 --node_rank=0 --nproc_per_node=4 --use_env train.py
Here command is for single node, 4 gpu. Tested only for single node.
You can watch loss graphs and samples in tensorboard by running tensorboard command in log folder.

Testing/Inference

Download pretrained model from this link(165 MB) in trained_checkpoint folder.
Put input images in input_images folder
Run python infer.py for inference.
Output will be saved in output_images

OR

Inference in colab from here

Acknowledgements

U2net model is from original u2net repo. Thanks to Xuebin Qin for amazing repo.
Complete repo follows structure of Pix2pixHD repo

cloth-segmentation's People

Contributors

Stargazers

Watchers

Forkers

minushuang sqella louixlouis ml-inory ericmiao0817 charlie839242 masom9768 zishanqin stoensin kirtinikam dmitriybartoshik nanlulu dongzhang89 pavan-1-2 brandon-eh mtkshu danailliu basemdabbour vinbert0203 beyonehan fmcnish-co andrej1a ruoque dawin2015 saranyan brandnewx marklr begyalcin mbrukman thiagoneves lukashedegaard ohadsgithub ovshake chrizandrcouture erwannmillon ifeherva bsmnyk sesinko kmiksa tieck-it spacecamp karthickal alexander-moore zaiqiangwu diullei xiaoqingwang dianjiaogit sharda-tech nihit2809 454126389 weinbery lucafrancis amananiket zhuxiongwei24 enrilohm lucifer1702 zlkbzlkb usingsystem007 barrierml franter666 profmatjr dustin-nguyen-qil cyborgsquirrel prog-ape piyushgit011 degengames antonytm jason-snow gagbaghdas linhong00316 reviewexplorer richard8610 deokgyuhan beetroit radiance-ai linecode jan06chen anjus1313 afei027 asutosh7hota psykeron ax-sh-ai soeminemi tejaswininikumbh tinaa23 hkercn xiankgx ramtri3d nhp-research

cloth-segmentation's Issues

How can upper body and Lower body segmentation can be separated out?

Hi,
Thanks for sharing this amazing work. Would like to know, once I have generated the output results, how I can separate out the upper body and lower body clothes from the image and process these two individual object for further task? Also, is their a JSON file that is being generated as output specifying the object detected in the image?

Please help.
Thanks

Is it possible to segregate the final output into each seperate label and have its original colour from the input image?

Model isn't accurate

Contour of another class is added. Even on high quality pictures. I've changed colours' labels but it has same bug also in original colours.

Prorosal: Pack model for Huggingface inference

Clothes segmentation is an important task in the field of computer vision and has various applications such as fashion analysis, virtual try-on, and image editing. However, there are currently limited pre-trained models available for clothes segmentation on the Hugging Face platform.

I propose adding a pre-trained clothes segmentation model to the Hugging Face platform.

您好！如何将分割结果输出为RGB彩色图？

多人图片的识别问题

发现在多人图片的识别效果很不好，如下图的右上角的结果。

然后我使用一下代码的处理方式（先通过detectronV2检测出人体框，然后将人体框逐个人体输入到网络中，最后进行融合）

def ClothSegMultiGen(self,img_cv,size=-1,):

        img = Image.fromarray(cv2.cvtColor(img_cv, cv2.COLOR_BGR2RGB))

        w,h = img.size

        body_boxes, _, sub_bodys = self.face_analysis.DetectronV2BodyBox(img_cv)
        total_rate = np.zeros((4,img_cv.shape[0],img_cv.shape[1]))-float('inf')  # 初始化一个数值最小的矩阵
        if len(sub_bodys)!=0:
            output_img  = np.zeros()
            for i in range(len(sub_bodys)):
                total_sub_rate = np.zeros((4,img_cv.shape[0],img_cv.shape[1]))-float('inf')  # 初始化一个数值最小的矩阵

                left, top, right, bottom = body_boxes[i][0],body_boxes[i][1],body_boxes[i][0]+body_boxes[i][2],body_boxes[i][1]+body_boxes[i][3]
                sub_img = img.crop((left, top, right, bottom))
                sub_img, sub_rate, sub_img_color   = self.ClothSegGen(sub_img,640)
                if i==0:

                    total_rate[:,top:bottom,left:right]= sub_rate
                else:
                    # np.argmax(sub_rate, axis=1)
                    total_sub_rate[:,top:bottom,left:right]= sub_rate
                    total_rate = maxTwoNumpy(total_sub_rate,total_rate)

            output_img = np.argmax(total_rate, axis=0)
            output_img_color = self.indexColor(output_img,w,h)
        else:
            output_img,_,output_img_color   = self.ClothSegGen(img,640)

def ClothSegGen(self,img_cv,size=-1,): #size表示短边长度
        img = Image.fromarray(cv2.cvtColor(img_cv, cv2.COLOR_BGR2RGB))
        w,h = img.size
        if size!=-1:
            if w>h:
                h_out = size
                w_out = w * h_out // h
            else:  
                w_out = size
                h_out = h * w_out // w

            img = img.resize((w_out,h_out))

        image_tensor = self.transform_rgb(img)
        image_tensor = torch.unsqueeze(image_tensor, 0)
        print('衣服识别时输入网络的图片大小：',image_tensor.shape)
        output_tensor = self.net(image_tensor.to(self.device))
        output_tensor = F.log_softmax(output_tensor[0], dim=1)

        output_tensor_ori = output_tensor.clone()
        output_tensor = torch.max(output_tensor_ori, dim=1, keepdim=True)[1]  # troch.max()[1]， 只返回最大值的每个索引


       
        output_tensor = torch.squeeze(output_tensor, dim=0)
        output_tensor = torch.squeeze(output_tensor, dim=0)
        output_arr = output_tensor.cpu().numpy()
        output_img = Image.fromarray(output_arr.astype("uint8"), mode="L")

        
        output_img_color = self.indexColor(output_img,w,h)



        # 单独处理出概率的数据
        output_tensor0 = output_tensor_ori.clone() # 但是此时的float并非0-1之间的概率值
        output_tensor0 = torch.squeeze(output_tensor0, dim=0) 
        # output_tensor0 = torch.squeeze(output_tensor0, dim=0)
        output_rate = output_tensor0.cpu().numpy() # 4*h*w
 

        return output_img,output_rate,output_img_color

但是发现这种方式有很大的融合问题（上一个人体框会压到下一个上面，比如上图中下面那张图片左边两个人连接处的问题）。不知道是不是因为log_softmax的问题？
因为以前是sigmoid的到的概率值，用同样的融合方式都能很正确的融合。

Can pretrain model to 11 class ?

final_label = first_channel + second_channel * 2 + third_channel * 3 conflict_mask = (final_label <= 3).astype("uint8") final_label = (conflict_mask) * final_label + (1 - conflict_mask) * 1 target_tensor = torch.as_tensor(final_label, dtype=torch.int64)

Hi,
I am very puzzled about the meaning of this code in "data/aligned_dataset.py".

`final_label = first_channel + second_channel * 2 + third_channel * 3

    conflict_mask = (final_label <= 3).astype("uint8")

    final_label = (conflict_mask) * final_label + (1 - conflict_mask) * 1

    target_tensor = torch.as_tensor(final_label, dtype=torch.int64)`

if I want to add another label , is this OK?
`
final_label = first_channel + second_channel * 2 + third_channel * 3 + fourth_channel * 4

    conflict_mask = (final_label <= 4).astype("uint8") # TODO

    final_label = (conflict_mask) * final_label + (1 - conflict_mask) * 1

    target_tensor = torch.as_tensor(final_label, dtype=torch.int64)`

I would appreciate your response. Thank you.

AttributeError: 'NoneType' object has no attribute 'to'; Issue with colab

IsADirectoryError: [Errno 21] Is a directory: 'input_images/.ipynb_checkpoints'

IsADirectoryError Traceback (most recent call last)
in ()
58 pbar = tqdm(total=len(images_list))
59 for image_name in images_list:
---> 60 img = Image.open(os.path.join(image_dir, image_name)).convert('RGB')
61 img_size = img.size
62 img = img.resize((768, 768), Image.BICUBIC)

/usr/local/lib/python3.7/dist-packages/PIL/Image.py in open(fp, mode)
2841
2842 if filename:
-> 2843 fp = builtins.open(filename, "rb")
2844 exclusive_fp = True
2845

IsADirectoryError: [Errno 21] Is a directory: 'input_images/.ipynb_checkpoints'

Inference error on trained checkpoints

Hi,
I ran the training script as your instructions which worked very well thank you. However, I'm getting an error when attempting to use my newly trained weights.

I changed the line of infer.py as such:

checkpoint_path = os.path.join("trained_checkpoint", "cloth_segm_u2net_latest.pth")
to
checkpoint_path = "results/training_cloth_segm_u2net_exp1/checkpoints/itr_00100000_u2net.pth"

And I can see the file sizes of the checkpoints aren't the same:

original:
$ ls -al trained_checkpoint/cloth_segm_u2net_latest.pth
-rw-r--r-- 1 user user 176625341 Mar 12 21:23 trained_checkpoint/cloth_segm_u2net_latest.pth

newly trained:
$ ls -al results/training_cloth_segm_u2net_exp1/checkpoints/itr_00100000_u2net.pth
-rw-r--r-- 1 user user 176607205 Mar 14 09:09 results/training_cloth_segm_u2net_exp1/checkpoints/itr_00100000_u2net.pth

The error I'm getting seems to drop the stageX names from the layers in the state dict. Any ideas?

Traceback (most recent call last):
  File "/nas/nns/fashion_seg/cloth_segmentation/infer.py", line 60, in <module>
    net = load_checkpoint_mgpu(net, checkpoint_path)
  File "/nas/nns/fashion_seg/cloth_segmentation/utils/saving_utils.py", line 29, in load_checkpoint_mgpu
    model.load_state_dict(new_state_dict)
  File "/home/user/.local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1671, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for U2NET:
	Missing key(s) in state_dict: "stage1.rebnconvin.conv_s1.weight", "stage1.rebnconvin.conv_s1.bias", "stage1.rebnconvin.bn_s1.weight", "stage1.rebnconvin.bn_s1.bias", "stage1.rebnconvin.bn_s1.running_mean", "stage1.rebnconvin.bn_s1.running_var", "stage1.rebnconv1.conv_s1.weight", "stage1.rebnconv1.conv_s1.bias", "stage1.rebnconv1.bn_s1.weight", "stage1.rebnconv1.bn_s1.bias", "stage1.rebnconv1.bn_s1.running_mean", "stage1.rebnconv1.bn_s1.running_var", "stage1.rebnconv2.conv_s1.weight", "stage1.rebnconv2.conv_s1.bias", "stage1.rebnconv2.bn_s1.weight", "stage1.rebnconv2.bn_s1.bias", "stage1.rebnconv2.bn_s1.running_mean", "stage1.rebnconv2.bn_s1.running_var", "stage1.rebnconv3.conv_s1.weight", "stage1.rebnconv3.conv_s1.bias", "stage1.rebnconv3.bn_s1.weight", "stage1.rebnconv3.bn_s1.bias", "stage1.rebnconv3.bn_s1.running_mean", "stage1.rebnconv3.bn_s1.running_var", "stage1.rebnconv4.conv_s1.weight", "stage1.rebnconv4.conv_s1.bias", "stage1.rebnconv4.bn_s1.weight", "stage1.rebnconv4.bn_s1.bias", "stage1.rebnconv4.bn_s1.running_mean", "stage1.rebnconv4.bn_s1.running_var", "stage1.rebnconv5.conv_s1.weight", "stage1.rebnconv5.conv_s1.bias", "stage1.rebnconv5.bn_s1.weight", "stage1.rebnconv5.bn_s1.bias", "stage1.rebnconv5.bn_s1.running_mean", "stage1.rebnconv5.bn_s1.running_var", "stage1.rebnconv6.conv_s1.weight", "stage1.rebnconv6.conv_s1.bias", "stage1.rebnconv6.bn_s1.weight", "stage1.rebnconv6.bn_s1.bias", "stage1.rebnconv6.bn_s1.running_mean", "stage1.rebnconv6.bn_s1.running_var", "stage1.rebnconv7.conv_s1.weight", "stage1.rebnconv7.conv_s1.bias", "stage1.rebnconv7.bn_s1.weight", "stage1.rebnconv7.bn_s1.bias", "stage1.rebnconv7.bn_s1.running_mean", "stage1.rebnconv7.bn_s1.running_var", "stage1.rebnconv6d.conv_s1.weight", "stage1.rebnconv6d.conv_s1.bias", "stage1.rebnconv6d.bn_s1.weight", "stage1.rebnconv6d.bn_s1.bias", "stage1.rebnconv6d.bn_s1.running_mean", "stage1.rebnconv6d.bn_s1.running_var", "stage1.rebnconv5d.conv_s1.weight", "stage1.rebnconv5d.conv_s1.bias", "stage1.rebnconv5d.bn_s1.weight", "stage1.rebnconv5d.bn_s1.bias", "stage1.rebnconv5d.bn_s1.running_mean", "stage1.rebnconv5d.bn_s1.running_var", "stage1.rebnconv4d.conv_s1.weight", "stage1.rebnconv4d.conv_s1.bias", "stage1.rebnconv4d.bn_s1.weight", "stage1.rebnconv4d.bn_s1.bias", "stage1.rebnconv4d.bn_s1.running_mean", "stage1.rebnconv4d.bn_s1.running_var", "stage1.rebnconv3d.conv_s1.weight", "stage1.rebnconv3d.conv_s1.bias", "stage1.rebnconv3d.bn_s1.weight", "stage1.rebnconv3d.bn_s1.bias", "stage1.rebnconv3d.bn_s1.running_mean", "stage1.rebnconv3d.bn_s1.running_var", "stage1.rebnconv2d.conv_s1.weight", "stage1.rebnconv2d.conv_s1.bias", "stage1.rebnconv2d.bn_s1.weight", "stage1.rebnconv2d.bn_s1.bias", "stage1.rebnconv2d.bn_s1.running_mean", "stage1.rebnconv2d.bn_s1.running_var", "stage1.rebnconv1d.conv_s1.weight", "stage1.rebnconv1d.conv_s1.bias", "stage1.rebnconv1d.bn_s1.weight", "stage1.rebnconv1d.bn_s1.bias", "stage1.rebnconv1d.bn_s1.running_mean", "stage1.rebnconv1d.bn_s1.running_var", "stage2.rebnconvin.conv_s1.weight", "stage2.rebnconvin.conv_s1.bias", "stage2.rebnconvin.bn_s1.weight", "stage2.rebnconvin.bn_s1.bias", "stage2.rebnconvin.bn_s1.running_mean", "stage2.rebnconvin.bn_s1.running_var", "stage2.rebnconv1.conv_s1.weight", "stage2.rebnconv1.conv_s1.bias", "stage2.rebnconv1.bn_s1.weight", "stage2.rebnconv1.bn_s1.bias", "stage2.rebnconv1.bn_s1.running_mean", "stage2.rebnconv1.bn_s1.running_var", "stage2.rebnconv2.conv_s1.weight", "stage2.rebnconv2.conv_s1.bias", "stage2.rebnconv2.bn_s1.weight", "stage2.rebnconv2.bn_s1.bias", "stage2.rebnconv2.bn_s1.running_mean", "stage2.rebnconv2.bn_s1.running_var", "stage2.rebnconv3.conv_s1.weight", "stage2.rebnconv3.conv_s1.bias", "stage2.rebnconv3.bn_s1.weight", "stage2.rebnconv3.bn_s1.bias", "stage2.rebnconv3.bn_s1.running_mean", "stage2.rebnconv3.bn_s1.running_var", "stage2.rebnconv4.conv_s1.weight", "stage2.rebnconv4.conv_s1.bias", "stage2.rebnconv4.bn_s1.weight", "stage2.rebnconv4.bn_s1.bias", "stage2.rebnconv4.bn_s1.running_mean", "stage2.rebnconv4.bn_s1.running_var", "stage2.rebnconv5.conv_s1.weight", "stage2.rebnconv5.conv_s1.bias", "stage2.rebnconv5.bn_s1.weight", "stage2.rebnconv5.bn_s1.bias", "stage2.rebnconv5.bn_s1.running_mean", "stage2.rebnconv5.bn_s1.running_var", "stage2.rebnconv6.conv_s1.weight", "stage2.rebnconv6.conv_s1.bias", "stage2.rebnconv6.bn_s1.weight", "stage2.rebnconv6.bn_s1.bias", "stage2.rebnconv6.bn_s1.running_mean", "stage2.rebnconv6.bn_s1.running_var", "stage2.rebnconv5d.conv_s1.weight", "stage2.rebnconv5d.conv_s1.bias", "stage2.rebnconv5d.bn_s1.weight", "stage2.rebnconv5d.bn_s1.bias", "stage2.rebnconv5d.bn_s1.running_mean", "stage2.rebnconv5d.bn_s1.running_var", "stage2.rebnconv4d.conv_s1.weight", "stage2.rebnconv4d.conv_s1.bias", "stage2.rebnconv4d.bn_s1.weight", "stage2.rebnconv4d.bn_s1.bias", "stage2.rebnconv4d.bn_s1.running_mean", "stage2.rebnconv4d.bn_s1.running_var", "stage2.rebnconv3d.conv_s1.weight", "stage2.rebnconv3d.conv_s1.bias", "stage2.rebnconv3d.bn_s1.weight", "stage2.rebnconv3d.bn_s1.bias", "stage2.rebnconv3d.bn_s1.running_mean", "stage2.rebnconv3d.bn_s1.running_var", "stage2.rebnconv2d.conv_s1.weight", "stage2.rebnconv2d.conv_s1.bias", "stage2.rebnconv2d.bn_s1.weight", "stage2.rebnconv2d.bn_s1.bias", "stage2.rebnconv2d.bn_s1.running_mean", "stage2.rebnconv2d.bn_s1.running_var", "stage2.rebnconv1d.conv_s1.weight", "stage2.rebnconv1d.conv_s1.bias", "stage2.rebnconv1d.bn_s1.weight", "stage2.rebnconv1d.bn_s1.bias", "stage2.rebnconv1d.bn_s1.running_mean", "stage2.rebnconv1d.bn_s1.running_var", "stage3.rebnconvin.conv_s1.weight", "stage3.rebnconvin.conv_s1.bias", "stage3.rebnconvin.bn_s1.weight", "stage3.rebnconvin.bn_s1.bias", "stage3.rebnconvin.bn_s1.running_mean", "stage3.rebnconvin.bn_s1.running_var", "stage3.rebnconv1.conv_s1.weight", "stage3.rebnconv1.conv_s1.bias", "stage3.rebnconv1.bn_s1.weight", "stage3.rebnconv1.bn_s1.bias", "stage3.rebnconv1.bn_s1.running_mean", "stage3.rebnconv1.bn_s1.running_var", "stage3.rebnconv2.conv_s1.weight", "stage3.rebnconv2.conv_s1.bias", "stage3.rebnconv2.bn_s1.weight", "stage3.rebnconv2.bn_s1.bias", "stage3.rebnconv2.bn_s1.running_mean", "stage3.rebnconv2.bn_s1.running_var", "stage3.rebnconv3.conv_s1.weight", "stage3.rebnconv3.conv_s1.bias", "stage3.rebnconv3.bn_s1.weight", "stage3.rebnconv3.bn_s1.bias", "stage3.rebnconv3.bn_s1.running_mean", "stage3.rebnconv3.bn_s1.running_var", "stage3.rebnconv4.conv_s1.weight", "stage3.rebnconv4.conv_s1.bias", "stage3.rebnconv4.bn_s1.weight", "stage3.rebnconv4.bn_s1.bias", "stage3.rebnconv4.bn_s1.running_mean", "stage3.rebnconv4.bn_s1.running_var", "stage3.rebnconv5.conv_s1.weight", "stage3.rebnconv5.conv_s1.bias", "stage3.rebnconv5.bn_s1.weight", "stage3.rebnconv5.bn_s1.bias", "stage3.rebnconv5.bn_s1.running_mean", "stage3.rebnconv5.bn_s1.running_var", "stage3.rebnconv4d.conv_s1.weight", "stage3.rebnconv4d.conv_s1.bias", "stage3.rebnconv4d.bn_s1.weight", "stage3.rebnconv4d.bn_s1.bias", "stage3.rebnconv4d.bn_s1.running_mean", "stage3.rebnconv4d.bn_s1.running_var", "stage3.rebnconv3d.conv_s1.weight", "stage3.rebnconv3d.conv_s1.bias", "stage3.rebnconv3d.bn_s1.weight", "stage3.rebnconv3d.bn_s1.bias", "stage3.rebnconv3d.bn_s1.running_mean", "stage3.rebnconv3d.bn_s1.running_var", "stage3.rebnconv2d.conv_s1.weight", "stage3.rebnconv2d.conv_s1.bias", "stage3.rebnconv2d.bn_s1.weight", "stage3.rebnconv2d.bn_s1.bias", "stage3.rebnconv2d.bn_s1.running_mean", "stage3.rebnconv2d.bn_s1.running_var", "stage3.rebnconv1d.conv_s1.weight", "stage3.rebnconv1d.conv_s1.bias", "stage3.rebnconv1d.bn_s1.weight", "stage3.rebnconv1d.bn_s1.bias", "stage3.rebnconv1d.bn_s1.running_mean", "stage3.rebnconv1d.bn_s1.running_var", "stage4.rebnconvin.conv_s1.weight", "stage4.rebnconvin.conv_s1.bias", "stage4.rebnconvin.bn_s1.weight", "stage4.rebnconvin.bn_s1.bias", "stage4.rebnconvin.bn_s1.running_mean", "stage4.rebnconvin.bn_s1.running_var", "stage4.rebnconv1.conv_s1.weight", "stage4.rebnconv1.conv_s1.bias", "stage4.rebnconv1.bn_s1.weight", "stage4.rebnconv1.bn_s1.bias", "stage4.rebnconv1.bn_s1.running_mean", "stage4.rebnconv1.bn_s1.running_var", "stage4.rebnconv2.conv_s1.weight", "stage4.rebnconv2.conv_s1.bias", "stage4.rebnconv2.bn_s1.weight", "stage4.rebnconv2.bn_s1.bias", "stage4.rebnconv2.bn_s1.running_mean", "stage4.rebnconv2.bn_s1.running_var", "stage4.rebnconv3.conv_s1.weight", "stage4.rebnconv3.conv_s1.bias", "stage4.rebnconv3.bn_s1.weight", "stage4.rebnconv3.bn_s1.bias", "stage4.rebnconv3.bn_s1.running_mean", "stage4.rebnconv3.bn_s1.running_var", "stage4.rebnconv4.conv_s1.weight", "stage4.rebnconv4.conv_s1.bias", "stage4.rebnconv4.bn_s1.weight", "stage4.rebnconv4.bn_s1.bias", "stage4.rebnconv4.bn_s1.running_mean", "stage4.rebnconv4.bn_s1.running_var", "stage4.rebnconv3d.conv_s1.weight", "stage4.rebnconv3d.conv_s1.bias", "stage4.rebnconv3d.bn_s1.weight", "stage4.rebnconv3d.bn_s1.bias", "stage4.rebnconv3d.bn_s1.running_mean", "stage4.rebnconv3d.bn_s1.running_var", "stage4.rebnconv2d.conv_s1.weight", "stage4.rebnconv2d.conv_s1.bias", "stage4.rebnconv2d.bn_s1.weight", "stage4.rebnconv2d.bn_s1.bias", "stage4.rebnconv2d.bn_s1.running_mean", "stage4.rebnconv2d.bn_s1.running_var", "stage4.rebnconv1d.conv_s1.weight", "stage4.rebnconv1d.conv_s1.bias", "stage4.rebnconv1d.bn_s1.weight", "stage4.rebnconv1d.bn_s1.bias", "stage4.rebnconv1d.bn_s1.running_mean", "stage4.rebnconv1d.bn_s1.running_var", "stage5.rebnconvin.conv_s1.weight", "stage5.rebnconvin.conv_s1.bias", "stage5.rebnconvin.bn_s1.weight", "stage5.rebnconvin.bn_s1.bias", "stage5.rebnconvin.bn_s1.running_mean", "stage5.rebnconvin.bn_s1.running_var", "stage5.rebnconv1.conv_s1.weight", "stage5.rebnconv1.conv_s1.bias", "stage5.rebnconv1.bn_s1.weight", "stage5.rebnconv1.bn_s1.bias", "stage5.rebnconv1.bn_s1.running_mean", "stage5.rebnconv1.bn_s1.running_var", "stage5.rebnconv2.conv_s1.weight", "stage5.rebnconv2.conv_s1.bias", "stage5.rebnconv2.bn_s1.weight", "stage5.rebnconv2.bn_s1.bias", "stage5.rebnconv2.bn_s1.running_mean", "stage5.rebnconv2.bn_s1.running_var", "stage5.rebnconv3.conv_s1.weight", "stage5.rebnconv3.conv_s1.bias", "stage5.rebnconv3.bn_s1.weight", "stage5.rebnconv3.bn_s1.bias", "stage5.rebnconv3.bn_s1.running_mean", "stage5.rebnconv3.bn_s1.running_var", "stage5.rebnconv4.conv_s1.weight", "stage5.rebnconv4.conv_s1.bias", "stage5.rebnconv4.bn_s1.weight", "stage5.rebnconv4.bn_s1.bias", "stage5.rebnconv4.bn_s1.running_mean", "stage5.rebnconv4.bn_s1.running_var", "stage5.rebnconv3d.conv_s1.weight", "stage5.rebnconv3d.conv_s1.bias", "stage5.rebnconv3d.bn_s1.weight", "stage5.rebnconv3d.bn_s1.bias", "stage5.rebnconv3d.bn_s1.running_mean", "stage5.rebnconv3d.bn_s1.running_var", "stage5.rebnconv2d.conv_s1.weight", "stage5.rebnconv2d.conv_s1.bias", "stage5.rebnconv2d.bn_s1.weight", "stage5.rebnconv2d.bn_s1.bias", "stage5.rebnconv2d.bn_s1.running_mean", "stage5.rebnconv2d.bn_s1.running_var", "stage5.rebnconv1d.conv_s1.weight", "stage5.rebnconv1d.conv_s1.bias", "stage5.rebnconv1d.bn_s1.weight", "stage5.rebnconv1d.bn_s1.bias", "stage5.rebnconv1d.bn_s1.running_mean", "stage5.rebnconv1d.bn_s1.running_var", "stage6.rebnconvin.conv_s1.weight", "stage6.rebnconvin.conv_s1.bias", "stage6.rebnconvin.bn_s1.weight", "stage6.rebnconvin.bn_s1.bias", "stage6.rebnconvin.bn_s1.running_mean", "stage6.rebnconvin.bn_s1.running_var", "stage6.rebnconv1.conv_s1.weight", "stage6.rebnconv1.conv_s1.bias", "stage6.rebnconv1.bn_s1.weight", "stage6.rebnconv1.bn_s1.bias", "stage6.rebnconv1.bn_s1.running_mean", "stage6.rebnconv1.bn_s1.running_var", "stage6.rebnconv2.conv_s1.weight", "stage6.rebnconv2.conv_s1.bias", "stage6.rebnconv2.bn_s1.weight", "stage6.rebnconv2.bn_s1.bias", "stage6.rebnconv2.bn_s1.running_mean", "stage6.rebnconv2.bn_s1.running_var", "stage6.rebnconv3.conv_s1.weight", "stage6.rebnconv3.conv_s1.bias", "stage6.rebnconv3.bn_s1.weight", "stage6.rebnconv3.bn_s1.bias", "stage6.rebnconv3.bn_s1.running_mean", "stage6.rebnconv3.bn_s1.running_var", "stage6.rebnconv4.conv_s1.weight", "stage6.rebnconv4.conv_s1.bias", "stage6.rebnconv4.bn_s1.weight", "stage6.rebnconv4.bn_s1.bias", "stage6.rebnconv4.bn_s1.running_mean", "stage6.rebnconv4.bn_s1.running_var", "stage6.rebnconv3d.conv_s1.weight", "stage6.rebnconv3d.conv_s1.bias", "stage6.rebnconv3d.bn_s1.weight", "stage6.rebnconv3d.bn_s1.bias", "stage6.rebnconv3d.bn_s1.running_mean", "stage6.rebnconv3d.bn_s1.running_var", "stage6.rebnconv2d.conv_s1.weight", "stage6.rebnconv2d.conv_s1.bias", "stage6.rebnconv2d.bn_s1.weight", "stage6.rebnconv2d.bn_s1.bias", "stage6.rebnconv2d.bn_s1.running_mean", "stage6.rebnconv2d.bn_s1.running_var", "stage6.rebnconv1d.conv_s1.weight", "stage6.rebnconv1d.conv_s1.bias", "stage6.rebnconv1d.bn_s1.weight", "stage6.rebnconv1d.bn_s1.bias", "stage6.rebnconv1d.bn_s1.running_mean", "stage6.rebnconv1d.bn_s1.running_var", "stage5d.rebnconvin.conv_s1.weight", "stage5d.rebnconvin.conv_s1.bias", "stage5d.rebnconvin.bn_s1.weight", "stage5d.rebnconvin.bn_s1.bias", "stage5d.rebnconvin.bn_s1.running_mean", "stage5d.rebnconvin.bn_s1.running_var", "stage5d.rebnconv1.conv_s1.weight", "stage5d.rebnconv1.conv_s1.bias", "stage5d.rebnconv1.bn_s1.weight", "stage5d.rebnconv1.bn_s1.bias", "stage5d.rebnconv1.bn_s1.running_mean", "stage5d.rebnconv1.bn_s1.running_var", "stage5d.rebnconv2.conv_s1.weight", "stage5d.rebnconv2.conv_s1.bias", "stage5d.rebnconv2.bn_s1.weight", "stage5d.rebnconv2.bn_s1.bias", "stage5d.rebnconv2.bn_s1.running_mean", "stage5d.rebnconv2.bn_s1.running_var", "stage5d.rebnconv3.conv_s1.weight", "stage5d.rebnconv3.conv_s1.bias", "stage5d.rebnconv3.bn_s1.weight", "stage5d.rebnconv3.bn_s1.bias", "stage5d.rebnconv3.bn_s1.running_mean", "stage5d.rebnconv3.bn_s1.running_var", "stage5d.rebnconv4.conv_s1.weight", "stage5d.rebnconv4.conv_s1.bias", "stage5d.rebnconv4.bn_s1.weight", "stage5d.rebnconv4.bn_s1.bias", "stage5d.rebnconv4.bn_s1.running_mean", "stage5d.rebnconv4.bn_s1.running_var", "stage5d.rebnconv3d.conv_s1.weight", "stage5d.rebnconv3d.conv_s1.bias", "stage5d.rebnconv3d.bn_s1.weight", "stage5d.rebnconv3d.bn_s1.bias", "stage5d.rebnconv3d.bn_s1.running_mean", "stage5d.rebnconv3d.bn_s1.running_var", "stage5d.rebnconv2d.conv_s1.weight", "stage5d.rebnconv2d.conv_s1.bias", "stage5d.rebnconv2d.bn_s1.weight", "stage5d.rebnconv2d.bn_s1.bias", "stage5d.rebnconv2d.bn_s1.running_mean", "stage5d.rebnconv2d.bn_s1.running_var", "stage5d.rebnconv1d.conv_s1.weight", "stage5d.rebnconv1d.conv_s1.bias", "stage5d.rebnconv1d.bn_s1.weight", "stage5d.rebnconv1d.bn_s1.bias", "stage5d.rebnconv1d.bn_s1.running_mean", "stage5d.rebnconv1d.bn_s1.running_var", "stage4d.rebnconvin.conv_s1.weight", "stage4d.rebnconvin.conv_s1.bias", "stage4d.rebnconvin.bn_s1.weight", "stage4d.rebnconvin.bn_s1.bias", "stage4d.rebnconvin.bn_s1.running_mean", "stage4d.rebnconvin.bn_s1.running_var", "stage4d.rebnconv1.conv_s1.weight", "stage4d.rebnconv1.conv_s1.bias", "stage4d.rebnconv1.bn_s1.weight", "stage4d.rebnconv1.bn_s1.bias", "stage4d.rebnconv1.bn_s1.running_mean", "stage4d.rebnconv1.bn_s1.running_var", "stage4d.rebnconv2.conv_s1.weight", "stage4d.rebnconv2.conv_s1.bias", "stage4d.rebnconv2.bn_s1.weight", "stage4d.rebnconv2.bn_s1.bias", "stage4d.rebnconv2.bn_s1.running_mean", "stage4d.rebnconv2.bn_s1.running_var", "stage4d.rebnconv3.conv_s1.weight", "stage4d.rebnconv3.conv_s1.bias", "stage4d.rebnconv3.bn_s1.weight", "stage4d.rebnconv3.bn_s1.bias", "stage4d.rebnconv3.bn_s1.running_mean", "stage4d.rebnconv3.bn_s1.running_var", "stage4d.rebnconv4.conv_s1.weight", "stage4d.rebnconv4.conv_s1.bias", "stage4d.rebnconv4.bn_s1.weight", "stage4d.rebnconv4.bn_s1.bias", "stage4d.rebnconv4.bn_s1.running_mean", "stage4d.rebnconv4.bn_s1.running_var", "stage4d.rebnconv3d.conv_s1.weight", "stage4d.rebnconv3d.conv_s1.bias", "stage4d.rebnconv3d.bn_s1.weight", "stage4d.rebnconv3d.bn_s1.bias", "stage4d.rebnconv3d.bn_s1.running_mean", "stage4d.rebnconv3d.bn_s1.running_var", "stage4d.rebnconv2d.conv_s1.weight", "stage4d.rebnconv2d.conv_s1.bias", "stage4d.rebnconv2d.bn_s1.weight", "stage4d.rebnconv2d.bn_s1.bias", "stage4d.rebnconv2d.bn_s1.running_mean", "stage4d.rebnconv2d.bn_s1.running_var", "stage4d.rebnconv1d.conv_s1.weight", "stage4d.rebnconv1d.conv_s1.bias", "stage4d.rebnconv1d.bn_s1.weight", "stage4d.rebnconv1d.bn_s1.bias", "stage4d.rebnconv1d.bn_s1.running_mean", "stage4d.rebnconv1d.bn_s1.running_var", "stage3d.rebnconvin.conv_s1.weight", "stage3d.rebnconvin.conv_s1.bias", "stage3d.rebnconvin.bn_s1.weight", "stage3d.rebnconvin.bn_s1.bias", "stage3d.rebnconvin.bn_s1.running_mean", "stage3d.rebnconvin.bn_s1.running_var", "stage3d.rebnconv1.conv_s1.weight", "stage3d.rebnconv1.conv_s1.bias", "stage3d.rebnconv1.bn_s1.weight", "stage3d.rebnconv1.bn_s1.bias", "stage3d.rebnconv1.bn_s1.running_mean", "stage3d.rebnconv1.bn_s1.running_var", "stage3d.rebnconv2.conv_s1.weight", "stage3d.rebnconv2.conv_s1.bias", "stage3d.rebnconv2.bn_s1.weight", "stage3d.rebnconv2.bn_s1.bias", "stage3d.rebnconv2.bn_s1.running_mean", "stage3d.rebnconv2.bn_s1.running_var", "stage3d.rebnconv3.conv_s1.weight", "stage3d.rebnconv3.conv_s1.bias", "stage3d.rebnconv3.bn_s1.weight", "stage3d.rebnconv3.bn_s1.bias", "stage3d.rebnconv3.bn_s1.running_mean", "stage3d.rebnconv3.bn_s1.running_var", "stage3d.rebnconv4.conv_s1.weight", "stage3d.rebnconv4.conv_s1.bias", "stage3d.rebnconv4.bn_s1.weight", "stage3d.rebnconv4.bn_s1.bias", "stage3d.rebnconv4.bn_s1.running_mean", "stage3d.rebnconv4.bn_s1.running_var", "stage3d.rebnconv5.conv_s1.weight", "stage3d.rebnconv5.conv_s1.bias", "stage3d.rebnconv5.bn_s1.weight", "stage3d.rebnconv5.bn_s1.bias", "stage3d.rebnconv5.bn_s1.running_mean", "stage3d.rebnconv5.bn_s1.running_var", "stage3d.rebnconv4d.conv_s1.weight", "stage3d.rebnconv4d.conv_s1.bias", "stage3d.rebnconv4d.bn_s1.weight", "stage3d.rebnconv4d.bn_s1.bias", "stage3d.rebnconv4d.bn_s1.running_mean", "stage3d.rebnconv4d.bn_s1.running_var", "stage3d.rebnconv3d.conv_s1.weight", "stage3d.rebnconv3d.conv_s1.bias", "stage3d.rebnconv3d.bn_s1.weight", "stage3d.rebnconv3d.bn_s1.bias", "stage3d.rebnconv3d.bn_s1.running_mean", "stage3d.rebnconv3d.bn_s1.running_var", "stage3d.rebnconv2d.conv_s1.weight", "stage3d.rebnconv2d.conv_s1.bias", "stage3d.rebnconv2d.bn_s1.weight", "stage3d.rebnconv2d.bn_s1.bias", "stage3d.rebnconv2d.bn_s1.running_mean", "stage3d.rebnconv2d.bn_s1.running_var", "stage3d.rebnconv1d.conv_s1.weight", "stage3d.rebnconv1d.conv_s1.bias", "stage3d.rebnconv1d.bn_s1.weight", "stage3d.rebnconv1d.bn_s1.bias", "stage3d.rebnconv1d.bn_s1.running_mean", "stage3d.rebnconv1d.bn_s1.running_var", "stage2d.rebnconvin.conv_s1.weight", "stage2d.rebnconvin.conv_s1.bias", "stage2d.rebnconvin.bn_s1.weight", "stage2d.rebnconvin.bn_s1.bias", "stage2d.rebnconvin.bn_s1.running_mean", "stage2d.rebnconvin.bn_s1.running_var", "stage2d.rebnconv1.conv_s1.weight", "stage2d.rebnconv1.conv_s1.bias", "stage2d.rebnconv1.bn_s1.weight", "stage2d.rebnconv1.bn_s1.bias", "stage2d.rebnconv1.bn_s1.running_mean", "stage2d.rebnconv1.bn_s1.running_var", "stage2d.rebnconv2.conv_s1.weight", "stage2d.rebnconv2.conv_s1.bias", "stage2d.rebnconv2.bn_s1.weight", "stage2d.rebnconv2.bn_s1.bias", "stage2d.rebnconv2.bn_s1.running_mean", "stage2d.rebnconv2.bn_s1.running_var", "stage2d.rebnconv3.conv_s1.weight", "stage2d.rebnconv3.conv_s1.bias", "stage2d.rebnconv3.bn_s1.weight", "stage2d.rebnconv3.bn_s1.bias", "stage2d.rebnconv3.bn_s1.running_mean", "stage2d.rebnconv3.bn_s1.running_var", "stage2d.rebnconv4.conv_s1.weight", "stage2d.rebnconv4.conv_s1.bias", "stage2d.rebnconv4.bn_s1.weight", "stage2d.rebnconv4.bn_s1.bias", "stage2d.rebnconv4.bn_s1.running_mean", "stage2d.rebnconv4.bn_s1.running_var", "stage2d.rebnconv5.conv_s1.weight", "stage2d.rebnconv5.conv_s1.bias", "stage2d.rebnconv5.bn_s1.weight", "stage2d.rebnconv5.bn_s1.bias", "stage2d.rebnconv5.bn_s1.running_mean", "stage2d.rebnconv5.bn_s1.running_var", "stage2d.rebnconv6.conv_s1.weight", "stage2d.rebnconv6.conv_s1.bias", "stage2d.rebnconv6.bn_s1.weight", "stage2d.rebnconv6.bn_s1.bias", "stage2d.rebnconv6.bn_s1.running_mean", "stage2d.rebnconv6.bn_s1.running_var", "stage2d.rebnconv5d.conv_s1.weight", "stage2d.rebnconv5d.conv_s1.bias", "stage2d.rebnconv5d.bn_s1.weight", "stage2d.rebnconv5d.bn_s1.bias", "stage2d.rebnconv5d.bn_s1.running_mean", "stage2d.rebnconv5d.bn_s1.running_var", "stage2d.rebnconv4d.conv_s1.weight", "stage2d.rebnconv4d.conv_s1.bias", "stage2d.rebnconv4d.bn_s1.weight", "stage2d.rebnconv4d.bn_s1.bias", "stage2d.rebnconv4d.bn_s1.running_mean", "stage2d.rebnconv4d.bn_s1.running_var", "stage2d.rebnconv3d.conv_s1.weight", "stage2d.rebnconv3d.conv_s1.bias", "stage2d.rebnconv3d.bn_s1.weight", "stage2d.rebnconv3d.bn_s1.bias", "stage2d.rebnconv3d.bn_s1.running_mean", "stage2d.rebnconv3d.bn_s1.running_var", "stage2d.rebnconv2d.conv_s1.weight", "stage2d.rebnconv2d.conv_s1.bias", "stage2d.rebnconv2d.bn_s1.weight", "stage2d.rebnconv2d.bn_s1.bias", "stage2d.rebnconv2d.bn_s1.running_mean", "stage2d.rebnconv2d.bn_s1.running_var", "stage2d.rebnconv1d.conv_s1.weight", "stage2d.rebnconv1d.conv_s1.bias", "stage2d.rebnconv1d.bn_s1.weight", "stage2d.rebnconv1d.bn_s1.bias", "stage2d.rebnconv1d.bn_s1.running_mean", "stage2d.rebnconv1d.bn_s1.running_var", "stage1d.rebnconvin.conv_s1.weight", "stage1d.rebnconvin.conv_s1.bias", "stage1d.rebnconvin.bn_s1.weight", "stage1d.rebnconvin.bn_s1.bias", "stage1d.rebnconvin.bn_s1.running_mean", "stage1d.rebnconvin.bn_s1.running_var", "stage1d.rebnconv1.conv_s1.weight", "stage1d.rebnconv1.conv_s1.bias", "stage1d.rebnconv1.bn_s1.weight", "stage1d.rebnconv1.bn_s1.bias", "stage1d.rebnconv1.bn_s1.running_mean", "stage1d.rebnconv1.bn_s1.running_var", "stage1d.rebnconv2.conv_s1.weight", "stage1d.rebnconv2.conv_s1.bias", "stage1d.rebnconv2.bn_s1.weight", "stage1d.rebnconv2.bn_s1.bias", "stage1d.rebnconv2.bn_s1.running_mean", "stage1d.rebnconv2.bn_s1.running_var", "stage1d.rebnconv3.conv_s1.weight", "stage1d.rebnconv3.conv_s1.bias", "stage1d.rebnconv3.bn_s1.weight", "stage1d.rebnconv3.bn_s1.bias", "stage1d.rebnconv3.bn_s1.running_mean", "stage1d.rebnconv3.bn_s1.running_var", "stage1d.rebnconv4.conv_s1.weight", "stage1d.rebnconv4.conv_s1.bias", "stage1d.rebnconv4.bn_s1.weight", "stage1d.rebnconv4.bn_s1.bias", "stage1d.rebnconv4.bn_s1.running_mean", "stage1d.rebnconv4.bn_s1.running_var", "stage1d.rebnconv5.conv_s1.weight", "stage1d.rebnconv5.conv_s1.bias", "stage1d.rebnconv5.bn_s1.weight", "stage1d.rebnconv5.bn_s1.bias", "stage1d.rebnconv5.bn_s1.running_mean", "stage1d.rebnconv5.bn_s1.running_var", "stage1d.rebnconv6.conv_s1.weight", "stage1d.rebnconv6.conv_s1.bias", "stage1d.rebnconv6.bn_s1.weight", "stage1d.rebnconv6.bn_s1.bias", "stage1d.rebnconv6.bn_s1.running_mean", "stage1d.rebnconv6.bn_s1.running_var", "stage1d.rebnconv7.conv_s1.weight", "stage1d.rebnconv7.conv_s1.bias", "stage1d.rebnconv7.bn_s1.weight", "stage1d.rebnconv7.bn_s1.bias", "stage1d.rebnconv7.bn_s1.running_mean", "stage1d.rebnconv7.bn_s1.running_var", "stage1d.rebnconv6d.conv_s1.weight", "stage1d.rebnconv6d.conv_s1.bias", "stage1d.rebnconv6d.bn_s1.weight", "stage1d.rebnconv6d.bn_s1.bias", "stage1d.rebnconv6d.bn_s1.running_mean", "stage1d.rebnconv6d.bn_s1.running_var", "stage1d.rebnconv5d.conv_s1.weight", "stage1d.rebnconv5d.conv_s1.bias", "stage1d.rebnconv5d.bn_s1.weight", "stage1d.rebnconv5d.bn_s1.bias", "stage1d.rebnconv5d.bn_s1.running_mean", "stage1d.rebnconv5d.bn_s1.running_var", "stage1d.rebnconv4d.conv_s1.weight", "stage1d.rebnconv4d.conv_s1.bias", "stage1d.rebnconv4d.bn_s1.weight", "stage1d.rebnconv4d.bn_s1.bias", "stage1d.rebnconv4d.bn_s1.running_mean", "stage1d.rebnconv4d.bn_s1.running_var", "stage1d.rebnconv3d.conv_s1.weight", "stage1d.rebnconv3d.conv_s1.bias", "stage1d.rebnconv3d.bn_s1.weight", "stage1d.rebnconv3d.bn_s1.bias", "stage1d.rebnconv3d.bn_s1.running_mean", "stage1d.rebnconv3d.bn_s1.running_var", "stage1d.rebnconv2d.conv_s1.weight", "stage1d.rebnconv2d.conv_s1.bias", "stage1d.rebnconv2d.bn_s1.weight", "stage1d.rebnconv2d.bn_s1.bias", "stage1d.rebnconv2d.bn_s1.running_mean", "stage1d.rebnconv2d.bn_s1.running_var", "stage1d.rebnconv1d.conv_s1.weight", "stage1d.rebnconv1d.conv_s1.bias", "stage1d.rebnconv1d.bn_s1.weight", "stage1d.rebnconv1d.bn_s1.bias", "stage1d.rebnconv1d.bn_s1.running_mean", "stage1d.rebnconv1d.bn_s1.running_var", "side1.weight", "side1.bias", "side2.weight", "side2.bias", "side3.weight", "side3.bias", "side4.weight", "side4.bias", "side5.weight", "side5.bias", "side6.weight", "side6.bias", "outconv.weight", "outconv.bias". 
	Unexpected key(s) in state_dict: "rebnconvin.conv_s1.weight", "rebnconvin.conv_s1.bias", "rebnconvin.bn_s1.weight", "rebnconvin.bn_s1.bias", "rebnconvin.bn_s1.running_mean", "rebnconvin.bn_s1.running_var", "rebnconvin.bn_s1.num_batches_tracked", "rebnconv1.conv_s1.weight", "rebnconv1.conv_s1.bias", "rebnconv1.bn_s1.weight", "rebnconv1.bn_s1.bias", "rebnconv1.bn_s1.running_mean", "rebnconv1.bn_s1.running_var", "rebnconv1.bn_s1.num_batches_tracked", "rebnconv2.conv_s1.weight", "rebnconv2.conv_s1.bias", "rebnconv2.bn_s1.weight", "rebnconv2.bn_s1.bias", "rebnconv2.bn_s1.running_mean", "rebnconv2.bn_s1.running_var", "rebnconv2.bn_s1.num_batches_tracked", "rebnconv3.conv_s1.weight", "rebnconv3.conv_s1.bias", "rebnconv3.bn_s1.weight", "rebnconv3.bn_s1.bias", "rebnconv3.bn_s1.running_mean", "rebnconv3.bn_s1.running_var", "rebnconv3.bn_s1.num_batches_tracked", "rebnconv4.conv_s1.weight", "rebnconv4.conv_s1.bias", "rebnconv4.bn_s1.weight", "rebnconv4.bn_s1.bias", "rebnconv4.bn_s1.running_mean", "rebnconv4.bn_s1.running_var", "rebnconv4.bn_s1.num_batches_tracked", "rebnconv5.conv_s1.weight", "rebnconv5.conv_s1.bias", "rebnconv5.bn_s1.weight", "rebnconv5.bn_s1.bias", "rebnconv5.bn_s1.running_mean", "rebnconv5.bn_s1.running_var", "rebnconv5.bn_s1.num_batches_tracked", "rebnconv6.conv_s1.weight", "rebnconv6.conv_s1.bias", "rebnconv6.bn_s1.weight", "rebnconv6.bn_s1.bias", "rebnconv6.bn_s1.running_mean", "rebnconv6.bn_s1.running_var", "rebnconv6.bn_s1.num_batches_tracked", "rebnconv7.conv_s1.weight", "rebnconv7.conv_s1.bias", "rebnconv7.bn_s1.weight", "rebnconv7.bn_s1.bias", "rebnconv7.bn_s1.running_mean", "rebnconv7.bn_s1.running_var", "rebnconv7.bn_s1.num_batches_tracked", "rebnconv6d.conv_s1.weight", "rebnconv6d.conv_s1.bias", "rebnconv6d.bn_s1.weight", "rebnconv6d.bn_s1.bias", "rebnconv6d.bn_s1.running_mean", "rebnconv6d.bn_s1.running_var", "rebnconv6d.bn_s1.num_batches_tracked", "rebnconv5d.conv_s1.weight", "rebnconv5d.conv_s1.bias", "rebnconv5d.bn_s1.weight", "rebnconv5d.bn_s1.bias", "rebnconv5d.bn_s1.running_mean", "rebnconv5d.bn_s1.running_var", "rebnconv5d.bn_s1.num_batches_tracked", "rebnconv4d.conv_s1.weight", "rebnconv4d.conv_s1.bias", "rebnconv4d.bn_s1.weight", "rebnconv4d.bn_s1.bias", "rebnconv4d.bn_s1.running_mean", "rebnconv4d.bn_s1.running_var", "rebnconv4d.bn_s1.num_batches_tracked", "rebnconv3d.conv_s1.weight", "rebnconv3d.conv_s1.bias", "rebnconv3d.bn_s1.weight", "rebnconv3d.bn_s1.bias", "rebnconv3d.bn_s1.running_mean", "rebnconv3d.bn_s1.running_var", "rebnconv3d.bn_s1.num_batches_tracked", "rebnconv2d.conv_s1.weight", "rebnconv2d.conv_s1.bias", "rebnconv2d.bn_s1.weight", "rebnconv2d.bn_s1.bias", "rebnconv2d.bn_s1.running_mean", "rebnconv2d.bn_s1.running_var", "rebnconv2d.bn_s1.num_batches_tracked", "rebnconv1d.conv_s1.weight", "rebnconv1d.conv_s1.bias", "rebnconv1d.bn_s1.weight", "rebnconv1d.bn_s1.bias", "rebnconv1d.bn_s1.running_mean", "rebnconv1d.bn_s1.running_var", "rebnconv1d.bn_s1.num_batches_tracked", ".rebnconvin.conv_s1.weight", ".rebnconvin.conv_s1.bias", ".rebnconvin.bn_s1.weight", ".rebnconvin.bn_s1.bias", ".rebnconvin.bn_s1.running_mean", ".rebnconvin.bn_s1.running_var", ".rebnconvin.bn_s1.num_batches_tracked", ".rebnconv1.conv_s1.weight", ".rebnconv1.conv_s1.bias", ".rebnconv1.bn_s1.weight", ".rebnconv1.bn_s1.bias", ".rebnconv1.bn_s1.running_mean", ".rebnconv1.bn_s1.running_var", ".rebnconv1.bn_s1.num_batches_tracked", ".rebnconv2.conv_s1.weight", ".rebnconv2.conv_s1.bias", ".rebnconv2.bn_s1.weight", ".rebnconv2.bn_s1.bias", ".rebnconv2.bn_s1.running_mean", ".rebnconv2.bn_s1.running_var", ".rebnconv2.bn_s1.num_batches_tracked", ".rebnconv3.conv_s1.weight", ".rebnconv3.conv_s1.bias", ".rebnconv3.bn_s1.weight", ".rebnconv3.bn_s1.bias", ".rebnconv3.bn_s1.running_mean", ".rebnconv3.bn_s1.running_var", ".rebnconv3.bn_s1.num_batches_tracked", ".rebnconv4.conv_s1.weight", ".rebnconv4.conv_s1.bias", ".rebnconv4.bn_s1.weight", ".rebnconv4.bn_s1.bias", ".rebnconv4.bn_s1.running_mean", ".rebnconv4.bn_s1.running_var", ".rebnconv4.bn_s1.num_batches_tracked", ".rebnconv3d.conv_s1.weight", ".rebnconv3d.conv_s1.bias", ".rebnconv3d.bn_s1.weight", ".rebnconv3d.bn_s1.bias", ".rebnconv3d.bn_s1.running_mean", ".rebnconv3d.bn_s1.running_var", ".rebnconv3d.bn_s1.num_batches_tracked", ".rebnconv2d.conv_s1.weight", ".rebnconv2d.conv_s1.bias", ".rebnconv2d.bn_s1.weight", ".rebnconv2d.bn_s1.bias", ".rebnconv2d.bn_s1.running_mean", ".rebnconv2d.bn_s1.running_var", ".rebnconv2d.bn_s1.num_batches_tracked", ".rebnconv1d.conv_s1.weight", ".rebnconv1d.conv_s1.bias", ".rebnconv1d.bn_s1.weight", ".rebnconv1d.bn_s1.bias", ".rebnconv1d.bn_s1.running_mean", ".rebnconv1d.bn_s1.running_var", ".rebnconv1d.bn_s1.num_batches_tracked", ".rebnconv5.conv_s1.weight", ".rebnconv5.conv_s1.bias", ".rebnconv5.bn_s1.weight", ".rebnconv5.bn_s1.bias", ".rebnconv5.bn_s1.running_mean", ".rebnconv5.bn_s1.running_var", ".rebnconv5.bn_s1.num_batches_tracked", ".rebnconv4d.conv_s1.weight", ".rebnconv4d.conv_s1.bias", ".rebnconv4d.bn_s1.weight", ".rebnconv4d.bn_s1.bias", ".rebnconv4d.bn_s1.running_mean", ".rebnconv4d.bn_s1.running_var", ".rebnconv4d.bn_s1.num_batches_tracked", ".rebnconv6.conv_s1.weight", ".rebnconv6.conv_s1.bias", ".rebnconv6.bn_s1.weight", ".rebnconv6.bn_s1.bias", ".rebnconv6.bn_s1.running_mean", ".rebnconv6.bn_s1.running_var", ".rebnconv6.bn_s1.num_batches_tracked", ".rebnconv5d.conv_s1.weight", ".rebnconv5d.conv_s1.bias", ".rebnconv5d.bn_s1.weight", ".rebnconv5d.bn_s1.bias", ".rebnconv5d.bn_s1.running_mean", ".rebnconv5d.bn_s1.running_var", ".rebnconv5d.bn_s1.num_batches_tracked", ".rebnconv7.conv_s1.weight", ".rebnconv7.conv_s1.bias", ".rebnconv7.bn_s1.weight", ".rebnconv7.bn_s1.bias", ".rebnconv7.bn_s1.running_mean", ".rebnconv7.bn_s1.running_var", ".rebnconv7.bn_s1.num_batches_tracked", ".rebnconv6d.conv_s1.weight", ".rebnconv6d.conv_s1.bias", ".rebnconv6d.bn_s1.weight", ".rebnconv6d.bn_s1.bias", ".rebnconv6d.bn_s1.running_mean", ".rebnconv6d.bn_s1.running_var", ".rebnconv6d.bn_s1.num_batches_tracked", "eight", "ias", ".weight", ".bias".

RuntimeError: CUDA out of memory.

Hello,
Thank you for providing the source code of this project.
I need to run the code on my own machine having 6 GB GPU.
When I am doing so, I am getting the following error:

RuntimeError: CUDA out of memory. Tried to allocate 12.00 MiB (GPU 0; 6.00 GiB total capacity; 5.30 GiB already allocated; 0 bytes free; 5.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

Can you please help me in solving this error?

Pretrained model file is not found using google drive link

same as title, google drive link of pre-trained model doesn't work. Could you upload a new one pre-trained model file?

[question] Segmentation mask dimensions

Hello there,
Is there a quick and easy way to get the model to return segmentation masks larger than a predetermined number of pixels?

Model very slow on cpu

Hi, thanks for excellent clothes segmentation model.
But model is very slow when inferencing on cpu. In colab demo it takes 14s. for one inference without GPU. Even after converting model to onnx, and openvino format I reached only 2s for inference.
Any ideas why it's so slow?

[bug] labels should start from 1

Very nice code!
I found a small bug that results in missing one cloth category (cape) and in confusing some cateegories a bit.

In data/aligned_dataset.py at line 82 you add 1 to each label:

labels.append(int(label) + 1)

However, in lines 121-123, you didn't do the same:

upperbody = [0, 1, 2, 3, 4, 5]
lowerbody = [6, 7, 8]
wholebody = [9, 10, 11, 12]

Hope it helps.

Best

TypeError: Caught TypeError in DataLoader worker process 0. and TypeError: 'float' object cannot be interpreted as an integer

Traceback (most recent call last):
File "C:\dl\u2net-cloth-segmentation-main\train.py", line 182, in
training_loop(opt)
File "C:\dl\u2net-cloth-segmentation-main\train.py", line 111, in training_loop
data_batch = next(get_data)
File "C:\dl\u2net-cloth-segmentation-main\data\custom_dataset_data_loader.py", line 50, in sample_data
for batch in loader:
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch\utils\data\dataloader.py", line 530, in next
data = self._next_data()
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch\utils\data\dataloader.py", line 1224, in _next_data
return self._process_data(data)
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch\utils\data\dataloader.py", line 1250, in _process_data
data.reraise()
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch_utils.py", line 457, in reraise
raise exception
TypeError: Caught TypeError in DataLoader worker process 0.
Original Traceback (most recent call last):
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch\utils\data_utils\worker.py", line 287, in _worker_loop
data = fetcher.fetch(index)
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch\utils\data_utils\fetch.py", line 49, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "C:\Anaconda3\envs\UnetTorch\lib\site-packages\torch\utils\data_utils\fetch.py", line 49, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "C:\dl\u2net-cloth-segmentation-main\data\aligned_dataset.py", line 74, in getitem
sub_mask = self.rle_decode(
File "C:\dl\u2net-cloth-segmentation-main\data\aligned_dataset.py", line 164, in rle_decode
img = np.zeros(shape[0] * shape[1], dtype=np.uint8)
TypeError: 'float' object cannot be interpreted as an integer

environment：
python = 3.9.0
pytorch =1.11.0

Cuda compilation tools, release 11.3, V11.3.58
Build cuda_11.3.r11.3/compiler.29745058_0

Why do I have this problem when I reproduce? Thank you

Possible to adjust sensitivity threshold?

Is it possible to adjust the threshold for detection so we can adjust the sensitivity of detection or is there no such variable in the code?

Getting the predictions

When I run the infer.py it segments the image but I don't get any of the detected objects or classes. and always have to refer to the image Is there a way I can get the segmented classes (e.g. upper, lower etc.) ?

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.