Upload 21 files

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 assets/compare_zoedepth.png filter=lfs diff=lfs merge=lfs -text
+dataset/splits/hypersim/train.txt filter=lfs diff=lfs merge=lfs -text

dataset/hypersim.py

@@ -0,0 +1,74 @@
+import cv2
+import h5py
+import numpy as np
+import torch
+from torch.utils.data import Dataset
+from torchvision.transforms import Compose
+
+from dataset.transform import Resize, NormalizeImage, PrepareForNet, Crop
+
+
+def hypersim_distance_to_depth(npyDistance):
+    intWidth, intHeight, fltFocal = 1024, 768, 886.81
+
+    npyImageplaneX = np.linspace((-0.5 * intWidth) + 0.5, (0.5 * intWidth) - 0.5, intWidth).reshape(
+        1, intWidth).repeat(intHeight, 0).astype(np.float32)[:, :, None]
+    npyImageplaneY = np.linspace((-0.5 * intHeight) + 0.5, (0.5 * intHeight) - 0.5,
+                                 intHeight).reshape(intHeight, 1).repeat(intWidth, 1).astype(np.float32)[:, :, None]
+    npyImageplaneZ = np.full([intHeight, intWidth, 1], fltFocal, np.float32)
+    npyImageplane = np.concatenate(
+        [npyImageplaneX, npyImageplaneY, npyImageplaneZ], 2)
+
+    npyDepth = npyDistance / np.linalg.norm(npyImageplane, 2, 2) * fltFocal
+    return npyDepth
+
+
+class Hypersim(Dataset):
+    def __init__(self, filelist_path, mode, size=(518, 518)):
+        
+        self.mode = mode
+        self.size = size
+        
+        with open(filelist_path, 'r') as f:
+            self.filelist = f.read().splitlines()
+        
+        net_w, net_h = size
+        self.transform = Compose([
+            Resize(
+                width=net_w,
+                height=net_h,
+                resize_target=True if mode == 'train' else False,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=14,
+                resize_method='lower_bound',
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
+            PrepareForNet(),
+        ] + ([Crop(size[0])] if self.mode == 'train' else []))
+        
+    def __getitem__(self, item):
+        img_path = self.filelist[item].split(' ')[0]
+        depth_path = self.filelist[item].split(' ')[1]
+        
+        image = cv2.imread(img_path)
+        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) / 255.0
+        
+        depth_fd = h5py.File(depth_path, "r")
+        distance_meters = np.array(depth_fd['dataset'])
+        depth = hypersim_distance_to_depth(distance_meters)
+        
+        sample = self.transform({'image': image, 'depth': depth})
+
+        sample['image'] = torch.from_numpy(sample['image'])
+        sample['depth'] = torch.from_numpy(sample['depth'])
+        
+        sample['valid_mask'] = (torch.isnan(sample['depth']) == 0)
+        sample['depth'][sample['valid_mask'] == 0] = 0
+        
+        sample['image_path'] = self.filelist[item].split(' ')[0]
+        
+        return sample
+
+    def __len__(self):
+        return len(self.filelist)

dataset/kitti.py

@@ -0,0 +1,57 @@
+import cv2
+import torch
+from torch.utils.data import Dataset
+from torchvision.transforms import Compose
+
+from dataset.transform import Resize, NormalizeImage, PrepareForNet
+
+
+class KITTI(Dataset):
+    def __init__(self, filelist_path, mode, size=(518, 518)):
+        if mode != 'val':
+            raise NotImplementedError
+        
+        self.mode = mode
+        self.size = size
+        
+        with open(filelist_path, 'r') as f:
+            self.filelist = f.read().splitlines()
+        
+        net_w, net_h = size
+        self.transform = Compose([
+            Resize(
+                width=net_w,
+                height=net_h,
+                resize_target=True if mode == 'train' else False,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=14,
+                resize_method='lower_bound',
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
+            PrepareForNet(),
+        ])
+    
+    def __getitem__(self, item):
+        img_path = self.filelist[item].split(' ')[0]
+        depth_path = self.filelist[item].split(' ')[1]
+        
+        image = cv2.imread(img_path)
+        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) / 255.0
+        
+        depth = cv2.imread(depth_path, cv2.IMREAD_UNCHANGED).astype('float32')
+        
+        sample = self.transform({'image': image, 'depth': depth})
+        
+        sample['image'] = torch.from_numpy(sample['image'])
+        sample['depth'] = torch.from_numpy(sample['depth'])
+        sample['depth'] = sample['depth'] / 256.0  # convert in meters
+        
+        sample['valid_mask'] = sample['depth'] > 0
+        
+        sample['image_path'] = self.filelist[item].split(' ')[0]
+        
+        return sample
+
+    def __len__(self):
+        return len(self.filelist)

dataset/pbr.py

@@ -0,0 +1,71 @@
+import cv2
+import torch
+from torch.utils.data import Dataset
+from torchvision.transforms import Compose
+
+from dataset.transform import Resize, NormalizeImage, PrepareForNet, Crop
+
+class PBRDataset(Dataset):
+    def __init__(self, filelist_path, mode, size=(512, 512)):
+        self.mode = mode
+        self.size = size
+        
+        # Read filelist using @@ as delimiter
+        self.filelist = []
+        with open(filelist_path, 'r') as f:
+            for line in f:
+                line = line.strip()
+                # Split on @@ delimiter
+                if '@@' in line:  # Use @@ as delimiter between paths
+                    parts = line.split('@@')
+                    if len(parts) == 2:
+                        self.filelist.append((parts[0].strip(), parts[1].strip()))
+        
+        print(f"Loaded {len(self.filelist)} image pairs")
+        
+        net_w, net_h = size
+        self.transform = Compose([
+            Resize(
+                width=net_w,
+                height=net_h,
+                resize_target=True if mode == 'train' else False,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=12,
+                resize_method='lower_bound',
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
+            PrepareForNet(),
+        ] + ([Crop(size[0])] if self.mode == 'train' else []))
+        
+    def __getitem__(self, item):
+        try:
+            img_path, disp_path = self.filelist[item]
+            
+            image = cv2.imread(img_path)
+            if image is None:
+                print(f"Failed to load image: {img_path}")
+                return self.__getitem__((item + 1) % len(self.filelist))
+            image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) / 255.0
+            
+            depth = cv2.imread(disp_path, cv2.IMREAD_GRAYSCALE)
+            if depth is None:
+                print(f"Failed to load depth: {disp_path}")
+                return self.__getitem__((item + 1) % len(self.filelist))
+            depth = depth.astype('float32') / 255.0
+
+            sample = self.transform({'image': image, 'depth': depth})
+
+            sample['image'] = torch.from_numpy(sample['image'])
+            sample['depth'] = torch.from_numpy(sample['depth'])
+            sample['valid_mask'] = torch.ones_like(sample['depth'], dtype=torch.bool)
+            sample['image_path'] = img_path
+            
+            return sample
+            
+        except Exception as e:
+            print(f"Error loading {item}: {str(e)}")
+            return self.__getitem__((item + 1) % len(self.filelist))
+
+    def __len__(self):
+        return len(self.filelist)

dataset/splits/hypersim/train.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f67054c519b4c008d7b58ada5735624780e5f89700bf07471747b3a1082b553
+size 13754433

dataset/splits/pbr/val.txt

@@ -0,0 +1,265 @@
+/mnt/f/Data/Test/valba/0002_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0002_particle_board_height.png
+/mnt/f/Data/Test/valba/0003_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0003_particle_board_height.png
+/mnt/f/Data/Test/valba/0004_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0004_particle_board_height.png
+/mnt/f/Data/Test/valba/0005_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0005_particle_board_height.png
+/mnt/f/Data/Test/valba/0006_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0006_particle_board_height.png
+/mnt/f/Data/Test/valba/0007_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0007_particle_board_height.png
+/mnt/f/Data/Test/valba/0009_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0009_particle_board_height.png
+/mnt/f/Data/Test/valba/0010_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0010_particle_board_height.png
+/mnt/f/Data/Test/valba/0011_particle_board_baked.png@@/mnt/f/Data/Test/valheight/0011_particle_board_height.png
+/mnt/f/Data/Test/valba/0012_shingles_baked.png@@/mnt/f/Data/Test/valheight/0012_shingles_height.png
+/mnt/f/Data/Test/valba/0013_shingles_baked.png@@/mnt/f/Data/Test/valheight/0013_shingles_height.png
+/mnt/f/Data/Test/valba/0014_shingles_baked.png@@/mnt/f/Data/Test/valheight/0014_shingles_height.png
+/mnt/f/Data/Test/valba/0015_shingles_baked.png@@/mnt/f/Data/Test/valheight/0015_shingles_height.png
+/mnt/f/Data/Test/valba/0016_shingles_baked.png@@/mnt/f/Data/Test/valheight/0016_shingles_height.png
+/mnt/f/Data/Test/valba/0017_shingles_baked.png@@/mnt/f/Data/Test/valheight/0017_shingles_height.png
+/mnt/f/Data/Test/valba/0018_shingles_baked.png@@/mnt/f/Data/Test/valheight/0018_shingles_height.png
+/mnt/f/Data/Test/valba/0019_shingles_baked.png@@/mnt/f/Data/Test/valheight/0019_shingles_height.png
+/mnt/f/Data/Test/valba/0021_shingles_baked.png@@/mnt/f/Data/Test/valheight/0021_shingles_height.png
+/mnt/f/Data/Test/valba/0022_shingles_baked.png@@/mnt/f/Data/Test/valheight/0022_shingles_height.png
+/mnt/f/Data/Test/valba/0024_shingles_baked.png@@/mnt/f/Data/Test/valheight/0024_shingles_height.png
+/mnt/f/Data/Test/valba/0025_shingles_baked.png@@/mnt/f/Data/Test/valheight/0025_shingles_height.png
+/mnt/f/Data/Test/valba/0026_shingles_baked.png@@/mnt/f/Data/Test/valheight/0026_shingles_height.png
+/mnt/f/Data/Test/valba/0027_shingles_baked.png@@/mnt/f/Data/Test/valheight/0027_shingles_height.png
+/mnt/f/Data/Test/valba/0028_shingles_baked.png@@/mnt/f/Data/Test/valheight/0028_shingles_height.png
+/mnt/f/Data/Test/valba/0029_shingles_baked.png@@/mnt/f/Data/Test/valheight/0029_shingles_height.png
+/mnt/f/Data/Test/valba/0030_shingles_baked.png@@/mnt/f/Data/Test/valheight/0030_shingles_height.png
+/mnt/f/Data/Test/valba/0031_shingles_baked.png@@/mnt/f/Data/Test/valheight/0031_shingles_height.png
+/mnt/f/Data/Test/valba/0032_shingles_baked.png@@/mnt/f/Data/Test/valheight/0032_shingles_height.png
+/mnt/f/Data/Test/valba/0033_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0033_organic_wood_height.png
+/mnt/f/Data/Test/valba/0034_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0034_organic_wood_height.png
+/mnt/f/Data/Test/valba/0035_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0035_organic_wood_height.png
+/mnt/f/Data/Test/valba/0036_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0036_organic_wood_height.png
+/mnt/f/Data/Test/valba/0037_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0037_organic_wood_height.png
+/mnt/f/Data/Test/valba/0038_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0038_organic_wood_height.png
+/mnt/f/Data/Test/valba/0039_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0039_organic_wood_height.png
+/mnt/f/Data/Test/valba/0040_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0040_organic_wood_height.png
+/mnt/f/Data/Test/valba/0041_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0041_organic_wood_height.png
+/mnt/f/Data/Test/valba/0042_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0042_organic_wood_height.png
+/mnt/f/Data/Test/valba/0043_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0043_organic_wood_height.png
+/mnt/f/Data/Test/valba/0044_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0044_organic_wood_height.png
+/mnt/f/Data/Test/valba/0045_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0045_organic_wood_height.png
+/mnt/f/Data/Test/valba/0046_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0046_organic_wood_height.png
+/mnt/f/Data/Test/valba/0047_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0047_organic_wood_height.png
+/mnt/f/Data/Test/valba/0049_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0049_organic_wood_height.png
+/mnt/f/Data/Test/valba/0050_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0050_organic_wood_height.png
+/mnt/f/Data/Test/valba/0051_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0051_organic_wood_height.png
+/mnt/f/Data/Test/valba/0052_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0052_organic_wood_height.png
+/mnt/f/Data/Test/valba/0053_organic_wood_baked.png@@/mnt/f/Data/Test/valheight/0053_organic_wood_height.png
+/mnt/f/Data/Test/valba/0055_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0055_cardboard_height.png
+/mnt/f/Data/Test/valba/0056_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0056_cardboard_height.png
+/mnt/f/Data/Test/valba/0057_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0057_cardboard_height.png
+/mnt/f/Data/Test/valba/0058_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0058_cardboard_height.png
+/mnt/f/Data/Test/valba/0059_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0059_cardboard_height.png
+/mnt/f/Data/Test/valba/0060_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0060_cardboard_height.png
+/mnt/f/Data/Test/valba/0061_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0061_cardboard_height.png
+/mnt/f/Data/Test/valba/0062_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0062_cardboard_height.png
+/mnt/f/Data/Test/valba/0063_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0063_cardboard_height.png
+/mnt/f/Data/Test/valba/0064_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0064_cardboard_height.png
+/mnt/f/Data/Test/valba/0065_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0065_cardboard_height.png
+/mnt/f/Data/Test/valba/0066_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0066_cardboard_height.png
+/mnt/f/Data/Test/valba/0067_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0067_cardboard_height.png
+/mnt/f/Data/Test/valba/0068_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0068_cardboard_height.png
+/mnt/f/Data/Test/valba/0070_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0070_cardboard_height.png
+/mnt/f/Data/Test/valba/0071_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0071_cardboard_height.png
+/mnt/f/Data/Test/valba/0072_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0072_cardboard_height.png
+/mnt/f/Data/Test/valba/0073_cardboard_baked.png@@/mnt/f/Data/Test/valheight/0073_cardboard_height.png
+/mnt/f/Data/Test/valba/0075_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0075_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0076_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0076_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0077_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0077_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0078_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0078_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0080_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0080_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0081_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0081_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0082_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0082_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0083_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0083_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0084_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0084_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0085_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0085_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0086_frog_leather_wrinkles_baked.png@@/mnt/f/Data/Test/valheight/0086_frog_leather_wrinkles_height.png
+/mnt/f/Data/Test/valba/0087_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0087_stone_floor_height.png
+/mnt/f/Data/Test/valba/0089_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0089_stone_floor_height.png
+/mnt/f/Data/Test/valba/0090_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0090_stone_floor_height.png
+/mnt/f/Data/Test/valba/0091_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0091_stone_floor_height.png
+/mnt/f/Data/Test/valba/0092_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0092_stone_floor_height.png
+/mnt/f/Data/Test/valba/0094_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0094_stone_floor_height.png
+/mnt/f/Data/Test/valba/0095_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0095_stone_floor_height.png
+/mnt/f/Data/Test/valba/0096_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0096_stone_floor_height.png
+/mnt/f/Data/Test/valba/0097_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0097_stone_floor_height.png
+/mnt/f/Data/Test/valba/0099_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0099_stone_floor_height.png
+/mnt/f/Data/Test/valba/0100_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0100_stone_floor_height.png
+/mnt/f/Data/Test/valba/0101_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0101_stone_floor_height.png
+/mnt/f/Data/Test/valba/0102_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0102_stone_floor_height.png
+/mnt/f/Data/Test/valba/0103_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0103_stone_floor_height.png
+/mnt/f/Data/Test/valba/0106_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0106_stone_floor_height.png
+/mnt/f/Data/Test/valba/0107_stone_floor_baked.png@@/mnt/f/Data/Test/valheight/0107_stone_floor_height.png
+/mnt/f/Data/Test/valba/0109_cliff_baked.png@@/mnt/f/Data/Test/valheight/0109_cliff_height.png
+/mnt/f/Data/Test/valba/0110_cliff_baked.png@@/mnt/f/Data/Test/valheight/0110_cliff_height.png
+/mnt/f/Data/Test/valba/0111_cliff_baked.png@@/mnt/f/Data/Test/valheight/0111_cliff_height.png
+/mnt/f/Data/Test/valba/0113_cliff_baked.png@@/mnt/f/Data/Test/valheight/0113_cliff_height.png
+/mnt/f/Data/Test/valba/0114_cliff_baked.png@@/mnt/f/Data/Test/valheight/0114_cliff_height.png
+/mnt/f/Data/Test/valba/0115_cliff_baked.png@@/mnt/f/Data/Test/valheight/0115_cliff_height.png
+/mnt/f/Data/Test/valba/0116_cliff_baked.png@@/mnt/f/Data/Test/valheight/0116_cliff_height.png
+/mnt/f/Data/Test/valba/0118_cliff_baked.png@@/mnt/f/Data/Test/valheight/0118_cliff_height.png
+/mnt/f/Data/Test/valba/0119_cliff_baked.png@@/mnt/f/Data/Test/valheight/0119_cliff_height.png
+/mnt/f/Data/Test/valba/0120_cliff_baked.png@@/mnt/f/Data/Test/valheight/0120_cliff_height.png
+/mnt/f/Data/Test/valba/0121_cliff_baked.png@@/mnt/f/Data/Test/valheight/0121_cliff_height.png
+/mnt/f/Data/Test/valba/0122_cliff_baked.png@@/mnt/f/Data/Test/valheight/0122_cliff_height.png
+/mnt/f/Data/Test/valba/0123_cliff_baked.png@@/mnt/f/Data/Test/valheight/0123_cliff_height.png
+/mnt/f/Data/Test/valba/0125_cliff_baked.png@@/mnt/f/Data/Test/valheight/0125_cliff_height.png
+/mnt/f/Data/Test/valba/0126_cliff_baked.png@@/mnt/f/Data/Test/valheight/0126_cliff_height.png
+/mnt/f/Data/Test/valba/0127_cliff_baked.png@@/mnt/f/Data/Test/valheight/0127_cliff_height.png
+/mnt/f/Data/Test/valba/0129_steel_003_bitmap_baked.png@@/mnt/f/Data/Test/valheight/0129_steel_003_bitmap_height.png
+/mnt/f/Data/Test/valba/0133_steel_003_bitmap_baked.png@@/mnt/f/Data/Test/valheight/0133_steel_003_bitmap_height.png
+/mnt/f/Data/Test/valba/0134_steel_003_bitmap_baked.png@@/mnt/f/Data/Test/valheight/0134_steel_003_bitmap_height.png
+/mnt/f/Data/Test/valba/0135_steel_003_bitmap_baked.png@@/mnt/f/Data/Test/valheight/0135_steel_003_bitmap_height.png
+/mnt/f/Data/Test/valba/0138_steel_003_bitmap_baked.png@@/mnt/f/Data/Test/valheight/0138_steel_003_bitmap_height.png
+/mnt/f/Data/Test/valba/0139_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0139_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0140_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0140_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0141_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0141_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0142_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0142_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0144_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0144_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0145_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0145_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0146_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0146_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0147_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0147_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0148_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0148_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0149_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0149_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0151_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0151_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0152_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0152_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0153_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0153_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0154_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0154_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0155_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0155_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0157_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0157_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0158_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0158_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0159_fabric_suit_vintage_baked.png@@/mnt/f/Data/Test/valheight/0159_fabric_suit_vintage_height.png
+/mnt/f/Data/Test/valba/0160_metal_bolted_square_plate_baked.png@@/mnt/f/Data/Test/valheight/0160_metal_bolted_square_plate_height.png
+/mnt/f/Data/Test/valba/0164_metal_bolted_square_plate_baked.png@@/mnt/f/Data/Test/valheight/0164_metal_bolted_square_plate_height.png
+/mnt/f/Data/Test/valba/0165_metal_bolted_square_plate_baked.png@@/mnt/f/Data/Test/valheight/0165_metal_bolted_square_plate_height.png
+/mnt/f/Data/Test/valba/0167_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0167_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0168_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0168_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0169_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0169_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0170_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0170_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0171_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0171_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0172_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0172_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0173_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0173_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0174_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0174_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0175_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0175_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0179_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0179_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0181_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0181_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0182_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0182_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0183_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0183_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0184_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0184_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0185_marco_b_broken_concrete_tiles_baked.png@@/mnt/f/Data/Test/valheight/0185_marco_b_broken_concrete_tiles_height.png
+/mnt/f/Data/Test/valba/0186_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0186_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0188_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0188_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0189_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0189_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0190_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0190_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0191_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0191_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0192_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0192_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0193_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0193_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0194_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0194_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0195_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0195_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0196_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0196_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0199_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0199_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0201_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0201_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0202_aluminium_brushed_baked.png@@/mnt/f/Data/Test/valheight/0202_aluminium_brushed_height.png
+/mnt/f/Data/Test/valba/0203_metal_aluminium_directional_baked.png@@/mnt/f/Data/Test/valheight/0203_metal_aluminium_directional_height.png
+/mnt/f/Data/Test/valba/0205_metal_aluminium_directional_baked.png@@/mnt/f/Data/Test/valheight/0205_metal_aluminium_directional_height.png
+/mnt/f/Data/Test/valba/0206_metal_aluminium_directional_baked.png@@/mnt/f/Data/Test/valheight/0206_metal_aluminium_directional_height.png
+/mnt/f/Data/Test/valba/0207_metal_aluminium_directional_baked.png@@/mnt/f/Data/Test/valheight/0207_metal_aluminium_directional_height.png
+/mnt/f/Data/Test/valba/0214_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0214_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0215_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0215_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0216_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0216_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0217_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0217_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0218_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0218_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0220_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0220_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0222_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0222_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0223_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0223_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0224_bark_old_ginko_baked.png@@/mnt/f/Data/Test/valheight/0224_bark_old_ginko_height.png
+/mnt/f/Data/Test/valba/0226_stucco_baked.png@@/mnt/f/Data/Test/valheight/0226_stucco_height.png
+/mnt/f/Data/Test/valba/0227_stucco_baked.png@@/mnt/f/Data/Test/valheight/0227_stucco_height.png
+/mnt/f/Data/Test/valba/0229_stucco_baked.png@@/mnt/f/Data/Test/valheight/0229_stucco_height.png
+/mnt/f/Data/Test/valba/0230_stucco_baked.png@@/mnt/f/Data/Test/valheight/0230_stucco_height.png
+/mnt/f/Data/Test/valba/0231_stucco_baked.png@@/mnt/f/Data/Test/valheight/0231_stucco_height.png
+/mnt/f/Data/Test/valba/0234_stucco_baked.png@@/mnt/f/Data/Test/valheight/0234_stucco_height.png
+/mnt/f/Data/Test/valba/0235_stucco_baked.png@@/mnt/f/Data/Test/valheight/0235_stucco_height.png
+/mnt/f/Data/Test/valba/0236_stucco_baked.png@@/mnt/f/Data/Test/valheight/0236_stucco_height.png
+/mnt/f/Data/Test/valba/0237_stucco_baked.png@@/mnt/f/Data/Test/valheight/0237_stucco_height.png
+/mnt/f/Data/Test/valba/0239_stucco_baked.png@@/mnt/f/Data/Test/valheight/0239_stucco_height.png
+/mnt/f/Data/Test/valba/0240_stucco_baked.png@@/mnt/f/Data/Test/valheight/0240_stucco_height.png
+/mnt/f/Data/Test/valba/0241_stucco_baked.png@@/mnt/f/Data/Test/valheight/0241_stucco_height.png
+/mnt/f/Data/Test/valba/0242_stucco_baked.png@@/mnt/f/Data/Test/valheight/0242_stucco_height.png
+/mnt/f/Data/Test/valba/0243_stucco_baked.png@@/mnt/f/Data/Test/valheight/0243_stucco_height.png
+/mnt/f/Data/Test/valba/0244_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0244_grass_paver_height.png
+/mnt/f/Data/Test/valba/0246_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0246_grass_paver_height.png
+/mnt/f/Data/Test/valba/0247_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0247_grass_paver_height.png
+/mnt/f/Data/Test/valba/0248_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0248_grass_paver_height.png
+/mnt/f/Data/Test/valba/0249_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0249_grass_paver_height.png
+/mnt/f/Data/Test/valba/0250_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0250_grass_paver_height.png
+/mnt/f/Data/Test/valba/0252_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0252_grass_paver_height.png
+/mnt/f/Data/Test/valba/0253_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0253_grass_paver_height.png
+/mnt/f/Data/Test/valba/0254_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0254_grass_paver_height.png
+/mnt/f/Data/Test/valba/0255_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0255_grass_paver_height.png
+/mnt/f/Data/Test/valba/0256_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0256_grass_paver_height.png
+/mnt/f/Data/Test/valba/0257_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0257_grass_paver_height.png
+/mnt/f/Data/Test/valba/0260_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0260_grass_paver_height.png
+/mnt/f/Data/Test/valba/0261_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0261_grass_paver_height.png
+/mnt/f/Data/Test/valba/0262_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0262_grass_paver_height.png
+/mnt/f/Data/Test/valba/0263_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0263_grass_paver_height.png
+/mnt/f/Data/Test/valba/0264_grass_paver_baked.png@@/mnt/f/Data/Test/valheight/0264_grass_paver_height.png
+/mnt/f/Data/Test/valba/0265_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0265_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0267_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0267_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0268_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0268_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0269_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0269_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0270_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0270_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0271_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0271_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0272_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0272_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0273_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0273_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0274_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0274_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0275_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0275_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0276_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0276_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0277_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0277_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0278_fabric_padded_wall_baked.png@@/mnt/f/Data/Test/valheight/0278_fabric_padded_wall_height.png
+/mnt/f/Data/Test/valba/0279_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0279_concrete_006_height.png
+/mnt/f/Data/Test/valba/0280_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0280_concrete_006_height.png
+/mnt/f/Data/Test/valba/0281_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0281_concrete_006_height.png
+/mnt/f/Data/Test/valba/0282_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0282_concrete_006_height.png
+/mnt/f/Data/Test/valba/0283_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0283_concrete_006_height.png
+/mnt/f/Data/Test/valba/0285_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0285_concrete_006_height.png
+/mnt/f/Data/Test/valba/0286_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0286_concrete_006_height.png
+/mnt/f/Data/Test/valba/0287_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0287_concrete_006_height.png
+/mnt/f/Data/Test/valba/0288_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0288_concrete_006_height.png
+/mnt/f/Data/Test/valba/0289_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0289_concrete_006_height.png
+/mnt/f/Data/Test/valba/0290_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0290_concrete_006_height.png
+/mnt/f/Data/Test/valba/0291_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0291_concrete_006_height.png
+/mnt/f/Data/Test/valba/0292_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0292_concrete_006_height.png
+/mnt/f/Data/Test/valba/0293_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0293_concrete_006_height.png
+/mnt/f/Data/Test/valba/0294_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0294_concrete_006_height.png
+/mnt/f/Data/Test/valba/0296_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0296_concrete_006_height.png
+/mnt/f/Data/Test/valba/0297_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0297_concrete_006_height.png
+/mnt/f/Data/Test/valba/0298_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0298_concrete_006_height.png
+/mnt/f/Data/Test/valba/0299_concrete_006_baked.png@@/mnt/f/Data/Test/valheight/0299_concrete_006_height.png
+/mnt/f/Data/Test/valba/0300_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0300_concrete_010_height.png
+/mnt/f/Data/Test/valba/0301_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0301_concrete_010_height.png
+/mnt/f/Data/Test/valba/0303_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0303_concrete_010_height.png
+/mnt/f/Data/Test/valba/0304_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0304_concrete_010_height.png
+/mnt/f/Data/Test/valba/0306_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0306_concrete_010_height.png
+/mnt/f/Data/Test/valba/0307_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0307_concrete_010_height.png
+/mnt/f/Data/Test/valba/0308_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0308_concrete_010_height.png
+/mnt/f/Data/Test/valba/0309_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0309_concrete_010_height.png
+/mnt/f/Data/Test/valba/0311_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0311_concrete_010_height.png
+/mnt/f/Data/Test/valba/0313_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0313_concrete_010_height.png
+/mnt/f/Data/Test/valba/0314_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0314_concrete_010_height.png
+/mnt/f/Data/Test/valba/0315_concrete_010_baked.png@@/mnt/f/Data/Test/valheight/0315_concrete_010_height.png
+/mnt/f/Data/Test/valba/0316_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0316_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0317_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0317_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0318_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0318_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0320_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0320_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0322_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0322_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0323_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0323_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0324_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0324_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0325_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0325_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0327_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0327_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0328_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0328_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/0331_marble_wall_01_baked.png@@/mnt/f/Data/Test/valheight/0331_marble_wall_01_height.png
+/mnt/f/Data/Test/valba/27_baked.png@@/mnt/f/Data/Test/valheight/27_height.png
+/mnt/f/Data/Test/valba/29_baked.png@@/mnt/f/Data/Test/valheight/29_height.png
+/mnt/f/Data/Test/valba/30_baked.png@@/mnt/f/Data/Test/valheight/30_height.png

dataset/transform.py

@@ -0,0 +1,277 @@
+import cv2
+import math
+import numpy as np
+import torch
+import torch.nn.functional as F
+
+
+def apply_min_size(sample, size, image_interpolation_method=cv2.INTER_AREA):
+    """Rezise the sample to ensure the given size. Keeps aspect ratio.
+
+    Args:
+        sample (dict): sample
+        size (tuple): image size
+
+    Returns:
+        tuple: new size
+    """
+    shape = list(sample["disparity"].shape)
+
+    if shape[0] >= size[0] and shape[1] >= size[1]:
+        return sample
+
+    scale = [0, 0]
+    scale[0] = size[0] / shape[0]
+    scale[1] = size[1] / shape[1]
+
+    scale = max(scale)
+
+    shape[0] = math.ceil(scale * shape[0])
+    shape[1] = math.ceil(scale * shape[1])
+
+    # resize
+    sample["image"] = cv2.resize(
+        sample["image"], tuple(shape[::-1]), interpolation=image_interpolation_method
+    )
+
+    sample["disparity"] = cv2.resize(
+        sample["disparity"], tuple(shape[::-1]), interpolation=cv2.INTER_NEAREST
+    )
+    sample["mask"] = cv2.resize(
+        sample["mask"].astype(np.float32),
+        tuple(shape[::-1]),
+        interpolation=cv2.INTER_NEAREST,
+    )
+    sample["mask"] = sample["mask"].astype(bool)
+
+    return tuple(shape)
+
+
+class Resize(object):
+    """Resize sample to given size (width, height).
+    """
+
+    def __init__(
+        self,
+        width,
+        height,
+        resize_target=True,
+        keep_aspect_ratio=False,
+        ensure_multiple_of=1,
+        resize_method="lower_bound",
+        image_interpolation_method=cv2.INTER_AREA,
+    ):
+        """Init.
+
+        Args:
+            width (int): desired output width
+            height (int): desired output height
+            resize_target (bool, optional):
+                True: Resize the full sample (image, mask, target).
+                False: Resize image only.
+                Defaults to True.
+            keep_aspect_ratio (bool, optional):
+                True: Keep the aspect ratio of the input sample.
+                Output sample might not have the given width and height, and
+                resize behaviour depends on the parameter 'resize_method'.
+                Defaults to False.
+            ensure_multiple_of (int, optional):
+                Output width and height is constrained to be multiple of this parameter.
+                Defaults to 1.
+            resize_method (str, optional):
+                "lower_bound": Output will be at least as large as the given size.
+                "upper_bound": Output will be at max as large as the given size. (Output size might be smaller than given size.)
+                "minimal": Scale as least as possible.  (Output size might be smaller than given size.)
+                Defaults to "lower_bound".
+        """
+        self.__width = width
+        self.__height = height
+
+        self.__resize_target = resize_target
+        self.__keep_aspect_ratio = keep_aspect_ratio
+        self.__multiple_of = ensure_multiple_of
+        self.__resize_method = resize_method
+        self.__image_interpolation_method = image_interpolation_method
+
+    def constrain_to_multiple_of(self, x, min_val=0, max_val=None):
+        y = (np.round(x / self.__multiple_of) * self.__multiple_of).astype(int)
+
+        if max_val is not None and y > max_val:
+            y = (np.floor(x / self.__multiple_of) * self.__multiple_of).astype(int)
+
+        if y < min_val:
+            y = (np.ceil(x / self.__multiple_of) * self.__multiple_of).astype(int)
+
+        return y
+
+    def get_size(self, width, height):
+        # determine new height and width
+        scale_height = self.__height / height
+        scale_width = self.__width / width
+
+        if self.__keep_aspect_ratio:
+            if self.__resize_method == "lower_bound":
+                # scale such that output size is lower bound
+                if scale_width > scale_height:
+                    # fit width
+                    scale_height = scale_width
+                else:
+                    # fit height
+                    scale_width = scale_height
+            elif self.__resize_method == "upper_bound":
+                # scale such that output size is upper bound
+                if scale_width < scale_height:
+                    # fit width
+                    scale_height = scale_width
+                else:
+                    # fit height
+                    scale_width = scale_height
+            elif self.__resize_method == "minimal":
+                # scale as least as possbile
+                if abs(1 - scale_width) < abs(1 - scale_height):
+                    # fit width
+                    scale_height = scale_width
+                else:
+                    # fit height
+                    scale_width = scale_height
+            else:
+                raise ValueError(
+                    f"resize_method {self.__resize_method} not implemented"
+                )
+
+        if self.__resize_method == "lower_bound":
+            new_height = self.constrain_to_multiple_of(
+                scale_height * height, min_val=self.__height
+            )
+            new_width = self.constrain_to_multiple_of(
+                scale_width * width, min_val=self.__width
+            )
+        elif self.__resize_method == "upper_bound":
+            new_height = self.constrain_to_multiple_of(
+                scale_height * height, max_val=self.__height
+            )
+            new_width = self.constrain_to_multiple_of(
+                scale_width * width, max_val=self.__width
+            )
+        elif self.__resize_method == "minimal":
+            new_height = self.constrain_to_multiple_of(scale_height * height)
+            new_width = self.constrain_to_multiple_of(scale_width * width)
+        else:
+            raise ValueError(f"resize_method {self.__resize_method} not implemented")
+
+        return (new_width, new_height)
+
+    def __call__(self, sample):
+        width, height = self.get_size(
+            sample["image"].shape[1], sample["image"].shape[0]
+        )
+
+        # resize sample
+        sample["image"] = cv2.resize(
+            sample["image"],
+            (width, height),
+            interpolation=self.__image_interpolation_method,
+        )
+
+        if self.__resize_target:
+            if "disparity" in sample:
+                sample["disparity"] = cv2.resize(
+                    sample["disparity"],
+                    (width, height),
+                    interpolation=cv2.INTER_NEAREST,
+                )
+
+            if "depth" in sample:
+                sample["depth"] = cv2.resize(
+                    sample["depth"], (width, height), interpolation=cv2.INTER_NEAREST
+                )
+
+            if "semseg_mask" in sample:
+                # sample["semseg_mask"] = cv2.resize(
+                #     sample["semseg_mask"], (width, height), interpolation=cv2.INTER_NEAREST
+                # )
+                sample["semseg_mask"] = F.interpolate(torch.from_numpy(sample["semseg_mask"]).float()[None, None, ...], (height, width), mode='nearest').numpy()[0, 0]
+                
+            if "mask" in sample:
+                sample["mask"] = cv2.resize(
+                    sample["mask"].astype(np.float32),
+                    (width, height),
+                    interpolation=cv2.INTER_NEAREST,
+                )
+                # sample["mask"] = sample["mask"].astype(bool)
+
+        # print(sample['image'].shape, sample['depth'].shape)
+        return sample
+
+
+class NormalizeImage(object):
+    """Normlize image by given mean and std.
+    """
+
+    def __init__(self, mean, std):
+        self.__mean = mean
+        self.__std = std
+
+    def __call__(self, sample):
+        sample["image"] = (sample["image"] - self.__mean) / self.__std
+
+        return sample
+
+
+class PrepareForNet(object):
+    """Prepare sample for usage as network input.
+    """
+
+    def __init__(self):
+        pass
+
+    def __call__(self, sample):
+        image = np.transpose(sample["image"], (2, 0, 1))
+        sample["image"] = np.ascontiguousarray(image).astype(np.float32)
+
+        if "mask" in sample:
+            sample["mask"] = sample["mask"].astype(np.float32)
+            sample["mask"] = np.ascontiguousarray(sample["mask"])
+        
+        if "depth" in sample:
+            depth = sample["depth"].astype(np.float32)
+            sample["depth"] = np.ascontiguousarray(depth)
+            
+        if "semseg_mask" in sample:
+            sample["semseg_mask"] = sample["semseg_mask"].astype(np.float32)
+            sample["semseg_mask"] = np.ascontiguousarray(sample["semseg_mask"])
+
+        return sample
+
+
+class Crop(object):
+    """Crop sample for batch-wise training. Image is of shape CxHxW
+    """
+
+    def __init__(self, size):
+        if isinstance(size, int):
+            self.size = (size, size)
+        else:
+            self.size = size
+
+    def __call__(self, sample):
+        h, w = sample['image'].shape[-2:]
+        assert h >= self.size[0] and w >= self.size[1], 'Wrong size'
+        
+        h_start = np.random.randint(0, h - self.size[0] + 1)
+        w_start = np.random.randint(0, w - self.size[1] + 1)
+        h_end = h_start + self.size[0]
+        w_end = w_start + self.size[1]
+        
+        sample['image'] = sample['image'][:, h_start: h_end, w_start: w_end]
+        
+        if "depth" in sample:
+            sample["depth"] = sample["depth"][h_start: h_end, w_start: w_end]
+        
+        if "mask" in sample:
+            sample["mask"] = sample["mask"][h_start: h_end, w_start: w_end]
+            
+        if "semseg_mask" in sample:
+            sample["semseg_mask"] = sample["semseg_mask"][h_start: h_end, w_start: w_end]
+            
+        return sample

dataset/vkitti2.py

@@ -0,0 +1,54 @@
+import cv2
+import torch
+from torch.utils.data import Dataset
+from torchvision.transforms import Compose
+
+from dataset.transform import Resize, NormalizeImage, PrepareForNet, Crop
+
+
+class VKITTI2(Dataset):
+    def __init__(self, filelist_path, mode, size=(518, 518)):
+        
+        self.mode = mode
+        self.size = size
+        
+        with open(filelist_path, 'r') as f:
+            self.filelist = f.read().splitlines()
+        
+        net_w, net_h = size
+        self.transform = Compose([
+            Resize(
+                width=net_w,
+                height=net_h,
+                resize_target=True if mode == 'train' else False,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=14,
+                resize_method='lower_bound',
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
+            PrepareForNet(),
+        ] + ([Crop(size[0])] if self.mode == 'train' else []))
+    
+    def __getitem__(self, item):
+        img_path = self.filelist[item].split(' ')[0]
+        depth_path = self.filelist[item].split(' ')[1]
+        
+        image = cv2.imread(img_path)
+        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) / 255.0
+        
+        depth = cv2.imread(depth_path, cv2.IMREAD_ANYCOLOR | cv2.IMREAD_ANYDEPTH) / 100.0  # cm to m
+        
+        sample = self.transform({'image': image, 'depth': depth})
+
+        sample['image'] = torch.from_numpy(sample['image'])
+        sample['depth'] = torch.from_numpy(sample['depth'])
+        
+        sample['valid_mask'] = (sample['depth'] <= 80)
+        
+        sample['image_path'] = self.filelist[item].split(' ')[0]
+        
+        return sample
+
+    def __len__(self):
+        return len(self.filelist)