app.py

from huggingface_hub import from_pretrained_fastai
import gradio as gr

from fastai.vision.all import *

import torchvision.transforms as transforms
import torchvision.transforms as transforms

from fastai.basics import *
from fastai.vision import models
from fastai.vision.all import *
from fastai.metrics import *
from fastai.data.all import *
from fastai.callback import *
from pathlib import Path

import random
import PIL

#Definimos las funciones de transformacion que hemos creado en la practica para poder tratar los datos de entrada y que funcione bien
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
def transform_image(image):
    my_transforms = transforms.Compose([transforms.ToTensor(),
                                        transforms.Normalize(
                                            [0.485, 0.456, 0.406],
                                            [0.229, 0.224, 0.225])])
    image_aux = image
    return my_transforms(image_aux).unsqueeze(0).to(device)

class TargetMaskConvertTransform(ItemTransform):
    def __init__(self):
        pass
    def encodes(self, x):
        img,mask = x

        #Convertimos a array
        mask = np.array(mask)

        mask[(mask!=255) & (mask!=150) & (mask!=76) & (mask!=74) & (mask!=29) & (mask!=25)]=0
        mask[mask==255]=1
        mask[mask==150]=2
        mask[mask==76]=4
        mask[mask==74]=4
        mask[mask==29]=3
        mask[mask==25]=3

        # Back to PILMask
        mask = PILMask.create(mask)
        return img, mask

from albumentations import (
    Compose,
    OneOf,
    ElasticTransform,
    GridDistortion,
    OpticalDistortion,
    HorizontalFlip,
    Rotate,
    Transpose,
    CLAHE,
    ShiftScaleRotate
)

def get_y_fn (x):
    return Path(str(x).replace("Images","Labels").replace("color","gt").replace(".jpg",".png"))

class SegmentationAlbumentationsTransform(ItemTransform):
    split_idx = 0

    def __init__(self, aug):
        self.aug = aug

    def encodes(self, x):
        img,mask = x
        aug = self.aug(image=np.array(img), mask=np.array(mask))
        return PILImage.create(aug["image"]), PILMask.create(aug["mask"])

#Cargamos el modelo

repo_id = "jegilj/Practica3"
learn = from_pretrained_fastai(repo_id)
model = learn.model
model = model.cpu()


# Funcion de predicción
def predict(img_ruta):
    img = PIL.Image.fromarray(img_ruta)
    image = transforms.Resize((480,640))(img)
    tensor = transform_image(image=image)
    model.to(device)
    with torch.no_grad():
      outputs = model(tensor)
    
    outputs = torch.argmax(outputs,1)
    mask = np.array(outputs.cpu())
    mask[mask==1]=255
    mask[mask==2]=150
    mask[mask==3]=29
    mask[mask==4]=74
    mask = np.reshape(mask,(480,640))
    return Image.fromarray(mask.astype('uint8'))

    
# Creamos la interfaz y la lanzamos. 
gr.Interface(fn=predict, inputs=gr.inputs.Image(shape=(480, 640)), outputs=gr.inputs.Image(shape=(480, 640)), examples=['color_184.jpg','color_189.jpg']).launch(share=False)