Update app.py

app.py

@@ -1,39 +1,35 @@
 import gradio as gr
-import subprocess
-import os
-
-def crowd_counting(image):
-    # Save the uploaded image
-    image_path = "test/uploaded.jpg"
-    image.save(image_path)
-
-    # Run the crowd counting model using subprocess
-    
-    
-    command = "python3 detect.py --weights weights/crowdhuman_yolov5m.pt --source {} --head --project runs/output --exist-ok".format(image_path)
-    subprocess.run(command, shell=True)
-
-    # Read the total_boxes from the file
-    total_boxes_path = "runs/output/output.txt"
-    with open(total_boxes_path, "r") as f:
-        total_boxes = f.read()
-
-    # Get the output image
-    output_image = "runs/output/output.jpg"
-
-    # Return the output image and total_boxes
-    return output_image, total_boxes
-
-# Define the input and output interfaces
-inputs = gr.inputs.Image(type="pil", label="Input Image")
-outputs = [gr.outputs.Image(type="pil", label="Output Image"), gr.outputs.Textbox(label="Total (Head) Count")]
-
-# Define the title and description
-title = "Crowd Counting"
-description = "<div style='text-align: center;'>This is a crowd counting application that uses a deep learning model to count the number of heads in an image.<br>Made by HTX (Q3) </div>"
-
-# Create the Gradio interface without the flag button
-gradio_interface = gr.Interface(fn=crowd_counting, inputs=inputs, outputs=outputs, title=title, description=description, allow_flagging="never")
-
-# Run the Gradio interface
-gradio_interface.launch()
+from transformers import DetrImageProcessor, DetrForObjectDetection
+import torch
+from PIL import Image
+
+# Carregar o processador e o modelo
+processor = DetrImageProcessor.from_pretrained("facebook/detr-resnet-50")
+model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
+
+def detect_people(image):
+    # Pré-processar a imagem
+    inputs = processor(images=image, return_tensors="pt")
+
+    # Realizar a inferência
+    outputs = model(**inputs)
+
+    # Processar os resultados
+    target_sizes = torch.tensor([image.size[::-1]])
+    results = processor.post_process_object_detection(outputs, target_sizes=target_sizes)[0]
+
+    # Contar o número de pessoas detectadas
+    num_pessoas = sum([1 for score, label in zip(results["scores"], results["labels"]) if score > 0.9 and model.config.id2label[label.item()] == "person"])
+    return num_pessoas
+
+# Criar a interface do Gradio
+iface = gr.Interface(
+    fn=detect_people, 
+    inputs=gr.Image(type="pil"), 
+    outputs=gr.Textbox(),
+    title="Contador de Pessoas em Imagens",
+    description="Carregue uma imagem para contar o número de pessoas detectadas nela."
+)
+
+# Executar a aplicação
+iface.launch()