adiren7/FoodVisionVIT

1 2import gradio as gr3import os4import torch5 6from model import create_vit_model7from timeit import default_timer as timer8 9class_names = ["pizza", "steak", "sushi"]10 11vit , vit_transforms = create_vit_model()12 13vit.load_state_dict(torch.load(f="09_pretrained_vit_feature_extractor_pizza_steak_sushi_20_percent.pth",14                    map_location=torch.device("cpu")))15 16def predict(img):17 18  img_tranformed = vit_transforms(img).unsqueeze(0)19 20  start_time = timer()21  vit.eval()22  with torch.inference_mode():23    y_pred = vit(img_tranformed)24 25  pred_time = round(timer() - start_time  , 4)26  y_proba = torch.softmax(y_pred , dim =1)27 28  pred_dict = { class_names[i]:j for i, j in enumerate( y_proba[0]) }29 30  return pred_dict , pred_time31 32 33title = "FoodVision Mini 🍕🥩🍣"34description = "An VITfeature extractor computer vision model to classify images of food as pizza, steak or sushi."35article = "Created at [PyTorch Model Deployment]."36 37# Create examples list from "examples/" directory38example_list = [["examples/" + example] for example in os.listdir("examples")]39 40# Create the Gradio demo41demo = gr.Interface(fn=predict, # mapping function from input to output42                    inputs=gr.Image(type="pil"), # what are the inputs?43                    outputs=[gr.Label(num_top_classes=3, label="Predictions"), # what are the outputs?44                             gr.Number(label="Prediction time (s)")], # our fn has two outputs, therefore we have two outputs45                    examples=example_list,46                    title=title,47                    description=description,48                    article=article)49 50# Launch the demo!51demo.launch()52