Spaces:
Sleeping
Sleeping
import gradio as gr | |
import torch | |
from Unet import UNet | |
import torchvision | |
from torchvision.transforms import functional as f | |
import os | |
from timeit import default_timer as timer | |
device = 'cpu' | |
model = UNet(device=device, in_channels=3, num_classes=3) | |
model.load_state_dict(torch.load("./data/models/Unet_v1.pth", map_location=torch.device('cpu'))) | |
image_transforms = torchvision.transforms.Compose([ | |
torchvision.transforms.Resize(size=(128, 128)), | |
torchvision.transforms.ToTensor() | |
]) | |
def predict(img): | |
start_time = timer() | |
img_transformed = image_transforms(img).to(device) | |
model.eval() | |
with torch.inference_mode(): | |
y_logits = model(img_transformed.unsqueeze(dim=0)).squeeze(dim=0) | |
predicted_label = torch.argmax(y_logits, dim=0).to('cpu') | |
for i in range(3): | |
for j in range(128): | |
for z in range(128): | |
img_transformed[i][j][z] = predicted_label[j][z] | |
img_transformed = f.to_pil_image(img_transformed) | |
return img_transformed, round((timer() - start_time), 3) | |
title = "Animal Segmentation" | |
description = "An UNet* feature extractor computer vision model to segment animal in an image.\nModel works more precisely on an image that only contains just one animal." | |
article = "U-Net: Convolutional Networks for Biomedical Image Segmentation (https://arxiv.org/abs/1505.04597)" | |
example_list = [["examples/" + example] for example in os.listdir("examples")] | |
demo = gr.Interface(fn=predict, # mapping function from input to output | |
inputs=gr.Image(type="pil"), # what are the inputs? | |
outputs=[gr.Image(label="Segmentation"), # what are the outputs? | |
gr.Number(label="Prediction time (s)")], # our fn has two outputs, therefore we have two outputs | |
# Create examples list from "examples/" directory | |
examples=example_list, | |
title=title, | |
description=description, | |
article=article) | |
demo.launch() | |