simple UI

app.py

@@ -5,23 +5,37 @@ from inference import DepthPredictor, SegmentPredictor
 from utils import create_3d_obj, create_3d_pc, point_cloud
 import numpy as np
 
-
-def snap(image, video):
+def produce_depth_map(image):
     depth_predictor = DepthPredictor()
     depth_result = depth_predictor.predict(image)
-    rgb_gltf_path = create_3d_obj(np.array(image), depth_result, path='./rgb.gltf')
-    
+    return depth_result
+
+def produce_segmentation_map(image):
     segment_predictor = SegmentPredictor()
     sam_result = segment_predictor.predict(image)
-    fig = point_cloud(np.array(sam_result), depth_result)
+    return sam_result
+
+def produce_3d_reconstruction(image):
+    depth_predictor = DepthPredictor()
+    depth_result = depth_predictor.predict(image)
+    rgb_gltf_path = create_3d_obj(np.array(image), depth_result, path='./rgb.gltf')
+    return rgb_gltf_path
 
-    return [image, depth_result, sam_result, rgb_gltf_path, fig]#[depth_result, gltf_path, gltf_path]
+def produce_point_cloud(depth_map, segmentation_map):
+    return point_cloud(np.array(segmentation_map), depth_map)
 
+def snap(image, depth_map, segmentation_map):
+    depth_result = produce_depth_map(image) if depth_map else None
+    sam_result = produce_segmentation_map(image) if segmentation_map else None
+    rgb_gltf_path = produce_3d_reconstruction(image) if depth_map else None
+    point_cloud_fig = produce_point_cloud(depth_result, sam_result) if (segmentation_map and depth_map) else None
 
+    return [image, depth_result, sam_result, rgb_gltf_path, point_cloud_fig]
 demo = gr.Interface(
     snap,
     inputs=[gr.Image(source="webcam", tool=None, label="Input Image", type="pil"),
-            gr.Video(source="webcam")],
+            "checkbox",
+            "checkbox"],
     outputs=[gr.Image(label="RGB"),
              gr.Image(label="predicted depth"),
              gr.Image(label="predicted segmentation"),