Update app.py

app.py

@@ -3,47 +3,58 @@ from transformers import DPTFeatureExtractor, DPTForDepthEstimation
 import torch
 import numpy as np
 from PIL import Image
+import open3d as o3d
 from pathlib import Path
+import os
 
-# Load model and feature extractor
 feature_extractor = DPTFeatureExtractor.from_pretrained("Intel/dpt-large")
 model = DPTForDepthEstimation.from_pretrained("Intel/dpt-large")
-model.eval()
 
-def process_image(image):
-    # Chuẩn hóa ảnh đầu vào
+
+def process_image(image_path):
+    image_path = Path(image_path)
+    image_raw = Image.open(image_path)
+    image = image_raw.resize(
+        (800, int(800 * image_raw.size[1] / image_raw.size[0])),
+        Image.Resampling.LANCZOS,
+    )
+
+    # prepare image for the model
     encoding = feature_extractor(image, return_tensors="pt")
-    
-    # Forward qua model
+
+    # forward pass
     with torch.no_grad():
         outputs = model(**encoding)
         predicted_depth = outputs.predicted_depth
-    
-    # Resize output về đúng kích thước ảnh gốc
+
+    # interpolate to original size
     prediction = torch.nn.functional.interpolate(
         predicted_depth.unsqueeze(1),
-        size=image.size[::-1],  # (H, W)
+        size=image.size[::-1],
         mode="bicubic",
-        align_corners=False
+        align_corners=False,
     ).squeeze()
-    
-    # Chuyển thành ảnh uint8
     output = prediction.cpu().numpy()
-    formatted = (output * 255 / np.max(output)).astype('uint8')
-    img = Image.fromarray(formatted)
+    depth_image = (output * 255 / np.max(output)).astype("uint8")
+    img = Image.fromarray(depth_image)
     return [img]
 
-# Interface
-title = "Demo: Zero-shot Depth Estimation with DPT"
-description = "Intel's DPT: Dense Prediction Transformer for depth estimation from a single image."
+
+
+title = "Demo: zero-shot depth estimation with DPT + 3D Point Cloud"
+description = "This demo is a variation from the original <a href='https://huggingface.co/spaces/nielsr/dpt-depth-estimation' target='_blank'>DPT Demo</a>. It uses the DPT model to predict the depth of an image and then uses 3D Point Cloud to create a 3D object."
+examples = [["examples/" + img] for img in os.listdir("examples/")]
 
 iface = gr.Interface(
-    fn=process_image, 
-    inputs=gr.inputs.Image(type="pil", label="Input Image"),
-    outputs=predicted_depth,
+    fn=process_image,
+    inputs=[gr.Image(type="filepath", label="Input Image")],
+    outputs=[
+        gr.Image(label="predicted depth", type="pil"),
+    ],
     title=title,
     description=description,
-    allow_flagging="never"
+    examples=examples,
+    allow_flagging="never",
+    cache_examples=False,
 )
-
-iface.launch(debug=True)
+iface.launch(debug=True, show_api=False)