import gradio as gr from PIL import Image import requests import numpy as np import urllib.request from urllib.request import urlretrieve import PIL.Image import torchvision.transforms as T import fastai from fastai.vision import * from fastai.utils.mem import * class FeatureLoss(nn.Module): def __init__(self, m_feat, layer_ids, layer_wgts): super().__init__() self.m_feat = m_feat self.loss_features = [self.m_feat[i] for i in layer_ids] self.hooks = hook_outputs(self.loss_features, detach=False) self.wgts = layer_wgts self.metric_names = ['pixel',] + [f'feat_{i}' for i in range(len(layer_ids)) ] + [f'gram_{i}' for i in range(len(layer_ids))] def make_features(self, x, clone=False): self.m_feat(x) return [(o.clone() if clone else o) for o in self.hooks.stored] def forward(self, input, target): out_feat = self.make_features(target, clone=True) in_feat = self.make_features(input) self.feat_losses = [base_loss(input,target)] self.feat_losses += [base_loss(f_in, f_out)*w for f_in, f_out, w in zip(in_feat, out_feat, self.wgts)] self.feat_losses += [base_loss(gram_matrix(f_in), gram_matrix(f_out))*w**2 * 5e3 for f_in, f_out, w in zip(in_feat, out_feat, self.wgts)] self.metrics = dict(zip(self.metric_names, self.feat_losses)) return sum(self.feat_losses) def __del__(self): self.hooks.remove() MODEL_URL = "https://www.dropbox.com/s/rz9nt35um1agf5y/t10T.pkl?dl=1" urllib.request.urlretrieve(MODEL_URL, "t10T.pkl") path = Path(".") learn=load_learner(path, 't10T.pkl') urlretrieve("https://s.hdnux.com/photos/01/07/33/71/18726490/5/1200x0.jpg","soccer1.jpg") urlretrieve("https://media.okmagazine.com/brand-img/IEPXUdkY7/0x0/2015/06/celebrity-tattoos-16-splash.jpg","soccer2.jpg") urlretrieve("https://newsmeter.in/wp-content/uploads/2020/06/Ajay-Devgn-Tattoo.jpg","baseball.jpg") urlretrieve("https://www.lofficielusa.com/_next/image?url=https%3A%2F%2Fwww.datocms-assets.com%2F39109%2F1612780326-1612385929513872-most-shocking-celebrity-tattoos-harry-styles.jpg%3Fauto%3Dformat%252Ccompress%26cs%3Dsrgb&w=3840&q=75","baseball2.jpeg") sample_images = [["soccer1.jpg"], ["soccer2.jpg"], ["baseball.jpg"], ["baseball2.jpeg"]] def predict(input): size = input.size img_t = T.ToTensor()(input) img_fast = Image(img_t) p,img_hr,b = learn.predict(img_fast) x = np.minimum(np.maximum(image2np(img_hr.data*255), 0), 255).astype(np.uint8) img = PIL.Image.fromarray(x) im1 = img.resize(size) return im1 gr_interface = gr.Interface(fn=predict, inputs=gr.Image(type="pil"), outputs="image", title='Skin-Deep',examples=sample_images).launch();