agrobot_base/Treinamento/models/ruas/process_Ruas.py

import torch
from PIL import Image
import torchvision.transforms as T
import matplotlib.pyplot as plt
import numpy as np
import cv2
import json

# Supondo que você tenha a estrutura do repositório e o módulo `network` conforme descrito no README
from network.modeling import deeplabv3plus_resnet50 as deeplabv3_model

# Configurações Iniciais
NUM_CLASSES = 4  # Pascal VOC possui 3 classes + 1 para o fundo
OUTPUT_STRIDE = 16  # Valor comum para DeepLab
MODEL_PATH = 'backup/ruasModel_final.pth'  # Caminho para o modelo pré-treinado

# Função para carregar o modelo
def load_model(model_path):
    model = deeplabv3_model(num_classes=NUM_CLASSES, output_stride=OUTPUT_STRIDE)
    model.load_state_dict(torch.load(model_path), strict=False)
    model.eval()  # Modo de avaliação
    return model

# Carregar o modelo
model = load_model(MODEL_PATH)

# Função para processar uma imagem e realizar a segmentação
def segment_image(image_path, model):
    image = Image.open(image_path).convert("RGB")
    transform = T.Compose([
        T.Resize(520),
        T.ToTensor(),
        T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
    ])
    
    input_tensor = transform(image).unsqueeze(0)
    with torch.no_grad():
        output = model(input_tensor)
    output_predictions = output.max(1)[1].squeeze().detach().cpu().numpy()
    return output_predictions

def read_labelmap(path):
    label_colors = []
    class_names = []
    with open(path, 'r') as file:
        for line in file.readlines():
            # Ignora linhas comentadas
            if line.startswith('#'):
                continue
            parts = line.strip().split(':')
            if len(parts) >= 2:
                label = parts[0].strip()
                color = tuple(map(int, parts[1].split(',')))
                class_names.append(label)
                label_colors.append(color)
    return np.array(label_colors), class_names

def decode_segmap(image, labelmap_path="dataset/labelmap.txt"):
    label_colors, class_names = read_labelmap(labelmap_path)
    nc = len(label_colors)
    
    detected_classes = set(np.unique(image))  # Conjunto das classes detectadas na imagem
    
    r = np.zeros_like(image).astype(np.uint8)
    g = np.zeros_like(image).astype(np.uint8)
    b = np.zeros_like(image).astype(np.uint8)
    
    for l in detected_classes:  # Usar apenas as classes detectadas
        if l < nc:  # Verifica se o índice está dentro do intervalo das cores definidas
            idx = image == l
            r[idx] = label_colors[l, 0]
            g[idx] = label_colors[l, 1]
            b[idx] = label_colors[l, 2]
        
    rgb = np.stack([r, g, b], axis=2)
    return rgb, class_names, label_colors, detected_classes

def display_segmentation(input_image_path, output_predictions):
    original_image = Image.open(input_image_path)
    original_image = np.array(original_image)
    
    seg_image, class_names, label_colors, detected_classes = decode_segmap(output_predictions)
    
    seg_image_resized = Image.fromarray(seg_image).resize((original_image.shape[1], original_image.shape[0]), resample=Image.NEAREST)
    seg_image_resized = np.array(seg_image_resized)
    
    overlayed_img = original_image * 0.6 + seg_image_resized * 0.4
    overlayed_img = overlayed_img.astype(np.uint8)
    
    plt.figure(figsize=(20, 10))
    
    # Imagem com Overlay e Contornos
    ax2 = plt.subplot(1, 1, 1)
    ax2.imshow(overlayed_img)
    ax2.set_title('Imagem com Overlay e Contornos')
    ax2.axis('off')

    # Desenha os contornos e pontos sobre a imagem
    for l in detected_classes:
        if l < len(label_colors):
            mask = (output_predictions == l).astype(np.uint8) * 255
            contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
            for contour in contours:
                contour = contour.squeeze()
                ax2.plot(contour[:, 0], contour[:, 1], color=np.array(label_colors[l])/255.0, linewidth=2)
    
    import matplotlib.patches as mpatches
    patches = [mpatches.Patch(color=np.array(label_colors[i])/255.0, label=class_names[i]) for i in detected_classes]
    plt.figlegend(handles=patches, loc='lower center', ncol=5, labelspacing=0.)
    
    plt.subplots_adjust(bottom=0.25)  # Ajuste conforme necessário para acomodar a legenda
    plt.show()


def generate_and_save_json(image, output_path='contours_data.json', labelmap_path="dataset/labelmap.txt"):
    label_colors, class_names = read_labelmap(labelmap_path)
    detected_classes = set(np.unique(image))
    contours_info = []

    for l in detected_classes:
        if l < len(label_colors):  # Verifica se o índice está dentro do intervalo das cores definidas
            # Extrai a máscara para a classe atual
            mask = (image == l).astype(np.uint8) * 255
            contours, _ = cv2.findContours(mask, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
            for contour in contours:
                # Calcula os pontos extremos e intermediários para o contorno atual
                contour_data = {
                    'Classe': class_names[l],
                    'Contornos': contour.squeeze().tolist()
                }
                contours_info.append(contour_data)
    
    # Salva os dados em um arquivo JSON
    with open(output_path, 'w') as f:
        json.dump(contours_info, f)

    print(f"Contour data saved to {output_path}")
    

image_path = 'dataset/images/81.jpeg'
# Exemplo de como usar a função `segment_image`
output_predictions = segment_image(image_path, model)
# Gera e salva os dados dos contornos em um arquivo JSON
generate_and_save_json(output_predictions)
# Chamada da função
display_segmentation(image_path, output_predictions)