import gradio as gr from PIL import Image, ImageFilter import numpy as np import torch from torch.autograd import Variable from torchvision import transforms import torch.nn.functional as F import gdown import os os.system("git clone https://github.com/xuebinqin/DIS") os.system("mv DIS/IS-Net/* .") # project imports from data_loader_cache import normalize, im_reader, im_preprocess from models import * #Helpers device = 'cuda' if torch.cuda.is_available() else 'cpu' # Download official weights if not os.path.exists("saved_models"): os.mkdir("saved_models") MODEL_PATH_URL = "https://drive.google.com/uc?id=1KyMpRjewZdyYfxHPYcd-ZbanIXtin0Sn" gdown.download(MODEL_PATH_URL, "saved_models/isnet.pth", use_cookies=False) class GOSNormalize(object): ''' Normalize the Image using torch.transforms ''' def __init__(self, mean=[0.485,0.456,0.406], std=[0.229,0.224,0.225]): self.mean = mean self.std = std def __call__(self,image): image = normalize(image,self.mean,self.std) return image transform = transforms.Compose([GOSNormalize([0.5,0.5,0.5],[1.0,1.0,1.0])]) def load_image(im_path, hypar): im = im_reader(im_path) im, im_shp = im_preprocess(im, hypar["cache_size"]) im = torch.divide(im,255.0) shape = torch.from_numpy(np.array(im_shp)) return transform(im).unsqueeze(0), shape.unsqueeze(0) # make a batch of image, shape def build_model(hypar,device): net = hypar["model"]#GOSNETINC(3,1) # convert to half precision if(hypar["model_digit"]=="half"): net.half() for layer in net.modules(): if isinstance(layer, nn.BatchNorm2d): layer.float() net.to(device) if(hypar["restore_model"]!=""): net.load_state_dict(torch.load(hypar["model_path"]+"/"+hypar["restore_model"], map_location=device)) net.to(device) net.eval() return net def predict(net, inputs_val, shapes_val, hypar, device): ''' Given an Image, predict the mask ''' net.eval() if(hypar["model_digit"]=="full"): inputs_val = inputs_val.type(torch.FloatTensor) else: inputs_val = inputs_val.type(torch.HalfTensor) inputs_val_v = Variable(inputs_val, requires_grad=False).to(device) # wrap inputs in Variable ds_val = net(inputs_val_v)[0] # list of 6 results pred_val = ds_val[0][0,:,:,:] # B x 1 x H x W # we want the first one which is the most accurate prediction ## recover the prediction spatial size to the orignal image size pred_val = torch.squeeze(F.upsample(torch.unsqueeze(pred_val,0),(shapes_val[0][0],shapes_val[0][1]),mode='bilinear')) ma = torch.max(pred_val) mi = torch.min(pred_val) pred_val = (pred_val-mi)/(ma-mi) # max = 1 if device == 'cuda': torch.cuda.empty_cache() return (pred_val.detach().cpu().numpy()*255).astype(np.uint8) # it is the mask we need # Set Parameters hypar = {} # paramters for inferencing hypar["model_path"] ="./saved_models" ## load trained weights from this path hypar["restore_model"] = "isnet.pth" ## name of the to-be-loaded weights hypar["interm_sup"] = False ## indicate if activate intermediate feature supervision ## choose floating point accuracy -- hypar["model_digit"] = "full" ## indicates "half" or "full" accuracy of float number hypar["seed"] = 0 hypar["cache_size"] = [1024, 1024] ## cached input spatial resolution, can be configured into different size ## data augmentation parameters --- hypar["input_size"] = [1024, 1024] ## mdoel input spatial size, usually use the same value hypar["cache_size"], which means we don't further resize the images hypar["crop_size"] = [1024, 1024] ## random crop size from the input, it is usually set as smaller than hypar["cache_size"], e.g., [920,920] for data augmentation hypar["model"] = ISNetDIS() # Build Model net = build_model(hypar, device) def infer_mask(image: Image): image_path = image image_tensor, orig_size = load_image(image_path, hypar) mask = predict(net, image_tensor, orig_size, hypar, device) return Image.fromarray(mask).convert("L") def blur(image_set: list, blur_amount: int): blurred_image = image_set[0].filter(ImageFilter.GaussianBlur(blur_amount)) return Image.composite(image_set[0], blurred_image, image_set[1]) with gr.Blocks() as interface: default_im = Image.open("newman.jpg").convert("RGB") default_mask = Image.open("newman_mask.jpg").convert("RGB") examples_list = [os.path.join(os.path.dirname(__file__), "newman.jpg"), os.path.join(os.path.dirname(__file__), "abbey.jpg"), os.path.join(os.path.dirname(__file__), "julia.jpg") ] current_images = gr.State([default_im, default_mask]) mask_toggle = gr.State(False) gr.Markdown( """ ### Intelligent Photo Blur Using Dichotomous Image Segmentation This app leverages the machine learning engine built by Xuebin Qin ([https://github.com/xuebinqin/DIS](https://github.com/xuebinqin/DIS)) to mask the prominent subject within a photograph. The mask is used to keep the subject in clear focus while an adjustable slider is available to interactively blur the background. To use, upload a photo and press the run button. You can adjust the level of blur through the slider and view the mask using the "Show Generated Mask" button. """ ) with gr.Row(): with gr.Column(): input_image = gr.Image(value=default_im, type='filepath') run_button = gr.Button() gr.Examples(inputs=input_image, examples=examples_list) with gr.Column(): output_image = gr.Image() blur_slider = gr.Slider(0, 16, 5, step=1, label="Blur Amount") mask_button = gr.Button(value="Show Generated Mask") mask_image = gr.Image(value=default_mask, visible=False) def run(image: Image, current_images: gr.State): im_rgb = Image.open(image).convert("RGB") mask = infer_mask(image) return ( blur([im_rgb, mask], 5), mask, [im_rgb, mask] ) def reset_slider(): return gr.update(value=5) def show_mask(mask_toggle: gr.State): if mask_toggle == True: return gr.update(visible=False) else: return gr.update(visible=True) def toggle_mask(mask_toggle: gr.State): if mask_toggle == True: return False else: return True run_button.click(run, [input_image, current_images], [output_image, mask_image, current_images]) run_button.click(reset_slider, outputs=blur_slider) blur_slider.change(blur, [current_images, blur_slider], output_image, show_progress=False) mask_button.click(show_mask, inputs=mask_toggle, outputs=mask_image) mask_button.click(toggle_mask, inputs=mask_toggle, outputs=mask_toggle) interface.launch()