1

.gitignore

@@ -1,4 +1,6 @@
 *.pth
 *.pt
 .gitignore
-*.glb
+*.glb
+output/*
+app_test.py

app.py

@@ -17,6 +17,8 @@ import copy
 from tqdm import tqdm
 import cv2
 from PIL import Image
+import os.path as path
+import sys 
 
 from dust3r.inference import inference
 from dust3r.model import AsymmetricCroCo3DStereo
@@ -32,16 +34,19 @@ import spaces
 from huggingface_hub import hf_hub_download
 pl.ion()
 
+HERE_PATH = path.normpath(path.dirname(__file__))  # noqa
+sys.path.insert(0, HERE_PATH)  # noqa
+
 # for gpu >= Ampere and pytorch >= 1.12
 torch.backends.cuda.matmul.allow_tf32 = True
 batch_size = 1
 
 tmpdirname = tempfile.mkdtemp(suffix='_align3r_gradio_demo')
 image_size = 512
-silent = True
+silent = False
 gradio_delete_cache = 7200
-
-hf_hub_download(repo_id="apple/DepthPro", filename='depth_pro.pt', local_dir='third_party/ml-depth-pro/checkpoints/')
+print(f'{HERE_PATH}/third_party/ml-depth-pro/checkpoints/')
+hf_hub_download(repo_id="apple/DepthPro", filename='depth_pro.pt', local_dir=f'{HERE_PATH}/third_party/ml-depth-pro/checkpoints/')
 
 class FileState:
     def __init__(self, outfile_name=None):
@@ -141,24 +146,16 @@ def local_get_reconstructed_scene(filelist, min_conf_thr, as_pointcloud, mask_sk
         loss = scene.compute_global_alignment(init='mst', niter=300, schedule='linear', lr=lr)
     # mode = GlobalAlignerMode.PairViewer
     # scene = global_aligner(output, device=device, mode=mode, verbose=not silent)
-    save_folder = './output'
+    
+    save_folder = './output/bear'
     os.makedirs(save_folder, exist_ok=True)
     outfile = get_3D_model_from_scene(save_folder, silent, scene, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size)
       
     return outfile
 
 
-def run_example(snapshot, matching_conf_thr, min_conf_thr, cam_size, as_pointcloud, shared_intrinsics, filelist, **kw):
-    return local_get_reconstructed_scene(filelist, cam_size, **kw)
-# filelist = ['/home/lipeng/ljh_code/Video_Depth_CVPR2025-main/Align3R/data/davis/DAVIS/JPEGImages/480p/bear/00000.jpg', '/home/lipeng/ljh_code/Video_Depth_CVPR2025-main/Align3R/data/davis/DAVIS/JPEGImages/480p/bear/00008.jpg','/home/lipeng/ljh_code/Video_Depth_CVPR2025-main/Align3R/data/davis/DAVIS/JPEGImages/480p/bear/00004.jpg', '/home/lipeng/ljh_code/Video_Depth_CVPR2025-main/Align3R/data/davis/DAVIS/JPEGImages/480p/bear/00010.jpg']
-# min_conf_thr = 1.1
-# as_pointcloud = True
-# mask_sky = False
-# clean_depth = True
-# transparent_cams = False
-# cam_size = 0.2
-# depth_prior_name = 'Depth Anything V2'
-# local_get_reconstructed_scene(filelist, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name)
+def run_example(snapshot, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, inputfiles, **kw):
+    return local_get_reconstructed_scene(inputfiles, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, **kw)
 
 css = """.gradio-container {margin: 0 !important; min-width: 100%};"""
 title = "Align3R Demo"
@@ -174,14 +171,14 @@ with gradio.Blocks(css=css, title=title, delete_cache=(gradio_delete_cache, grad
         snapshot = gradio.Image(None, visible=False)
         with gradio.Row():
             # adjust the camera size in the output pointcloud
-            cam_size = gradio.Slider(label="cam_size", value=0.2, minimum=0.001, maximum=1.0, step=0.001)
+            cam_size = gradio.Slider(label="cam_size", value=0.02, minimum=0.001, maximum=1.0, step=0.001)
 
             depth_prior_name = gradio.Dropdown(
             ["Depth Pro", "Depth Anything V2"], label="monocular depth estimation model", info="Select the monocular depth estimation model.")
-            min_conf_thr = gradio.Slider(label="min_conf_thr", value=1.1, minimum=0.0, maximum=20, step=0.01)
+            min_conf_thr = gradio.Slider(label="min_conf_thr", value=2, minimum=0.0, maximum=20, step=0.01)
         with gradio.Row():
             as_pointcloud = gradio.Checkbox(value=True, label="As pointcloud")
-            mask_sky = gradio.Checkbox(value=False, label="Mask sky")
+            mask_sky = gradio.Checkbox(value=True, label="Mask sky")
             clean_depth = gradio.Checkbox(value=True, label="Clean-up depthmaps")
             transparent_cams = gradio.Checkbox(value=False, label="Transparent cameras")
             # not to show camera
@@ -189,17 +186,74 @@ with gradio.Blocks(css=css, title=title, delete_cache=(gradio_delete_cache, grad
         run_btn = gradio.Button("Run")
         outmodel = gradio.Model3D()
 
-        # examples = gradio.Examples(
-        #     examples=[
-        #         ['./example/yellowman/frame_0003.png',
-        #             0.0, 1.5, 0.2, True, False,
-        #         ]
-        #     ],
-        #     inputs=[snapshot, matching_conf_thr, min_conf_thr, cam_size, as_pointcloud, shared_intrinsics, inputfiles],
-        #     outputs=[filestate, outmodel],
-        #     fn=run_example,
-        #     cache_examples="lazy",
-        # )
+        examples = gradio.Examples(
+            examples=[
+                [
+                    os.path.join(HERE_PATH, 'example/bear/00000.jpg'),
+                    2, True, True, True, False, 0.02, "Depth Anything V2",
+                     [os.path.join(HERE_PATH, 'example/bear/00000.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00001.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00002.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00003.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00004.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00005.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00006.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00007.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00008.jpg'),
+                      os.path.join(HERE_PATH, 'example/bear/00009.jpg'),
+                      ]
+                ],
+                [
+                    os.path.join(HERE_PATH, 'example/breakdance/00000.jpg'),
+                    2, True, True, True, False, 0.02, "Depth Anything V2",
+                     [os.path.join(HERE_PATH, 'example/breakdance/00000.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00001.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00002.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00003.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00004.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00005.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00006.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00007.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00008.jpg'),
+                      os.path.join(HERE_PATH, 'example/breakdance/00009.jpg'),
+                      ]
+                ],
+                [
+                    os.path.join(HERE_PATH, 'example/tennis/00000.jpg'),
+                    2, True, True, True, False, 0.02, "Depth Anything V2",
+                     [os.path.join(HERE_PATH, 'example/tennis/00000.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00001.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00002.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00003.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00004.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00005.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00006.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00007.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00008.jpg'),
+                      os.path.join(HERE_PATH, 'example/tennis/00009.jpg'),
+                      ]
+                ],
+                [
+                    os.path.join(HERE_PATH, 'example/camel/00000.jpg'),
+                    2, True, True, True, False, 0.02, "Depth Anything V2",
+                     [os.path.join(HERE_PATH, 'example/camel/00000.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00001.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00002.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00003.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00004.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00005.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00006.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00007.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00008.jpg'),
+                      os.path.join(HERE_PATH, 'example/camel/00009.jpg'),
+                      ]
+                ],
+            ],
+            inputs=[snapshot, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, inputfiles],
+            outputs=[outmodel],
+            fn=run_example,
+            cache_examples="lazy",
+        )
 
         # events
         run_btn.click(fn=local_get_reconstructed_scene,