Moved pfm file reading into dataset utils (#6270)

* Moved pfm file reading into dataset utils

* Made _read_pfm private. Fixed doc format issues.

Author: TeodorPoncu

torchvision/datasets/_optical_flow.py

@@ -1,6 +1,5 @@
 import itertools
 import os
-import re
 from abc import ABC, abstractmethod
 from glob import glob
 from pathlib import Path
@@ -10,7 +9,7 @@
 from PIL import Image
 
 from ..io.image import _read_png_16
-from .utils import verify_str_arg
+from .utils import verify_str_arg, _read_pfm
 from .vision import VisionDataset
 
 
@@ -472,31 +471,3 @@ def _read_16bits_png_with_flow_and_valid_mask(file_name):
 
     # For consistency with other datasets, we convert to numpy
     return flow.numpy(), valid_flow_mask.numpy()
-
-
-def _read_pfm(file_name):
-    """Read flow in .pfm format"""
-
-    with open(file_name, "rb") as f:
-        header = f.readline().rstrip()
-        if header != b"PF":
-            raise ValueError("Invalid PFM file")
-
-        dim_match = re.match(rb"^(\d+)\s(\d+)\s$", f.readline())
-        if not dim_match:
-            raise Exception("Malformed PFM header.")
-        w, h = (int(dim) for dim in dim_match.groups())
-
-        scale = float(f.readline().rstrip())
-        if scale < 0:  # little-endian
-            endian = "<"
-            scale = -scale
-        else:
-            endian = ">"  # big-endian
-
-        data = np.fromfile(f, dtype=endian + "f")
-
-    data = data.reshape(h, w, 3).transpose(2, 0, 1)
-    data = np.flip(data, axis=1)  # flip on h dimension
-    data = data[:2, :, :]
-    return data.astype(np.float32)

torchvision/datasets/utils.py

@@ -18,6 +18,7 @@
 from typing import Any, Callable, List, Iterable, Optional, TypeVar, Dict, IO, Tuple, Iterator
 from urllib.parse import urlparse
 
+import numpy as np
 import requests
 import torch
 from torch.utils.model_zoo import tqdm
@@ -483,3 +484,39 @@ def verify_str_arg(
         raise ValueError(msg)
 
     return value
+
+
+def _read_pfm(file_name: str, slice_channels: int = 2) -> np.ndarray:
+    """Read file in .pfm format. Might contain either 1 or 3 channels of data.
+
+    Args:
+        file_name (str): Path to the file.
+        slice_channels (int): Number of channels to slice out of the file.
+            Useful for reading different data formats stored in .pfm files: Optical Flows, Stereo Disparity Maps, etc.
+    """
+
+    with open(file_name, "rb") as f:
+        header = f.readline().rstrip()
+        if header not in [b"PF", b"Pf"]:
+            raise ValueError("Invalid PFM file")
+
+        dim_match = re.match(rb"^(\d+)\s(\d+)\s$", f.readline())
+        if not dim_match:
+            raise Exception("Malformed PFM header.")
+        w, h = (int(dim) for dim in dim_match.groups())
+
+        scale = float(f.readline().rstrip())
+        if scale < 0:  # little-endian
+            endian = "<"
+            scale = -scale
+        else:
+            endian = ">"  # big-endian
+
+        data = np.fromfile(f, dtype=endian + "f")
+
+    pfm_channels = 3 if header == b"PF" else 1
+
+    data = data.reshape(h, w, pfm_channels).transpose(2, 0, 1)
+    data = np.flip(data, axis=1)  # flip on h dimension
+    data = data[:slice_channels, :, :]
+    return data.astype(np.float32)