Fix Issue #1903 path variable

docs/source/user_guide/dataset_creation/datasets.rst

@@ -2225,6 +2225,7 @@ where `dataset.yaml` contains the following information:
 
 .. code-block:: text
 
+    path: <dataset_dir> # optional
     train: ./images/train/
     val: ./images/val/
 
@@ -2238,7 +2239,9 @@ See `this page <https://docs.ultralytics.com/tutorials/train-custom-datasets>`_
 for a full description of the possible format of `dataset.yaml`. In particular,
 the dataset may contain one or more splits with arbitrary names, as the
 specific split being imported or exported is specified by the `split` argument
-to :class:`fiftyone.utils.yolo.YOLOv5DatasetImporter`.
+to :class:`fiftyone.utils.yolo.YOLOv5DatasetImporter`. `dataset.yaml` can also
+be located outside of `<dataset_dir>` as long as the optional `path` is set to
+point to `<dataset_dir>`.
 
 .. note::
 
@@ -2259,7 +2262,9 @@ where `<target>` is the zero-based integer index of the object class label from
 `[0, 1] x [0, 1]`, and `<confidence>` is an optional detection confidence in
 `[0, 1]`.
 
-Unlabeled images have no corresponding TXT file in `labels/`.
+Unlabeled images have no corresponding TXT file in `labels/`. The label file
+path for each image is obtained by replacing `images/` with `labels/` in the
+respective image path.
 
 The image and labels directories for a given split may contain nested
 subfolders of parallelly organized images and labels.

docs/source/user_guide/export_datasets.rst

@@ -2192,6 +2192,7 @@ where `dataset.yaml` contains the following information:
 
 .. code-block:: text
 
+    path: <dataset_dir> # optional
     train: ./images/train/
     val: ./images/val/
 
@@ -2205,7 +2206,9 @@ See `this page <https://docs.ultralytics.com/tutorials/train-custom-datasets>`_
 for a full description of the possible format of `dataset.yaml`. In particular,
 the dataset may contain one or more splits with arbitrary names, as the
 specific split being imported or exported is specified by the `split` argument
-to :class:`fiftyone.utils.yolo.YOLOv5DatasetExporter`.
+to :class:`fiftyone.utils.yolo.YOLOv5DatasetExporter`. `dataset.yaml` can also
+be located outside of `<dataset_dir>` as long as the optional `path` is set to
+point to `<dataset_dir>`.
 
 The TXT files in `labels/` are space-delimited files where each row corresponds
 to an object in the image of the same name, in one of the following formats:
@@ -2221,7 +2224,9 @@ where `<target>` is the zero-based integer index of the object class label from
 be included only if you pass the optional `include_confidence=True` flag to
 the export.
 
-Unlabeled images have no corresponding TXT file in `labels/`.
+Unlabeled images have no corresponding TXT file in `labels/`. The label file
+path for each image is obtained by replacing `images/` with `labels/` in the
+respective image path.
 
 .. note::
 

fiftyone/utils/yolo.py

@@ -476,7 +476,8 @@ def setup(self):
                 % (self.yaml_path, self.split)
             )
 
-        data = d[self.split]
+        dataset_path = d.get("path", "")
+        data = os.path.join(dataset_path, d[self.split])
         classes = d.get("names", None)
 
         if etau.is_str(data) and data.endswith(".txt"):