Conditional dependencies

examples/load.ipynb

@@ -398,6 +398,20 @@
      "name": "#%%\n"
     }
    }
+  },
+  {
+   "cell_type": "markdown",
+   "source": [
+    "# WMT-SLT\n",
+    "\n",
+    "Instructions and example code are here: https://github.com/sign-language-processing/datasets/blob/master/sign_language_datasets/datasets/wmt_slt/README.md"
+   ],
+   "metadata": {
+    "collapsed": false,
+    "pycharm": {
+     "name": "#%% md\n"
+    }
+   }
   }
  ]
 }

requirements.txt

@@ -5,10 +5,8 @@ tfds-nightly
 tensorflow>=2.4.0
 numpy
 pytest
-mediapipe
 pytest-cov
 pympi-ling
 Pillow
 opencv-python==4.5.5.64
 requests
-srt

setup.py

@@ -11,14 +11,14 @@
 setup(
     name="sign-language-datasets",
     packages=packages,
-    version="0.0.9",
+    version="0.0.10",
     description="TFDS Datasets for sign language",
     author="Amit Moryossef",
     author_email="amitmoryossef@gmail.com",
     url="https://github.com/sign-language-processing/datasets",
     keywords=[],
     install_requires=["python-dotenv", "tqdm", "pose-format", "tfds-nightly", "tensorflow", "numpy", "pympi-ling",
-                      "Pillow", "opencv-python==4.5.5.64", "mediapipe", "srt"],
+                      "Pillow", "opencv-python==4.5.5.64"],
     tests_require=['pytest', 'pytest-cov'],
     long_description=long_description,
     long_description_content_type="text/markdown",

sign_language_datasets/datasets/wmt_slt/README.md

@@ -5,6 +5,11 @@ Until [merged by `tfds`](https://github.com/tensorflow/datasets/pull/3988), plea
 pip install git+ssh://git@github.com:AmitMY/datasets-1.git
 ```
 
+Then install conditional dependencies:
+
+```bash
+pip install srt mediapipe
+```
 
 ```py
 

sign_language_datasets/datasets/wmt_slt/utils.py

@@ -12,7 +12,6 @@
 import tempfile
 
 import numpy as np
-import mediapipe as mp
 
 from pose_format import Pose, PoseHeader
 from pose_format.numpy import NumPyPoseBody
@@ -21,10 +20,6 @@
 from pose_format.utils.holistic import holistic_components
 from pose_format.utils.openpose import load_frames_directory_dict
 
-mp_holistic = mp.solutions.holistic
-FACEMESH_CONTOURS_POINTS = [str(p) for p in
-                            sorted(set([p for p_tup in list(mp_holistic.FACEMESH_CONTOURS) for p in p_tup]))]
-
 
 def extract_tar_xz_file(filepath: str, target_dir: str):
     """
@@ -59,6 +54,19 @@ def read_openpose_surrey_format(filepath: str, fps: int, width: int, height: int
 
 
 def formatted_holistic_pose(width=1000, height=1000):
+
+    try:
+        import mediapipe as mp
+
+    except ImportError:
+        raise ImportError(
+            "Please install mediapipe with: pip install mediapipe"
+        )
+
+    mp_holistic = mp.solutions.holistic
+    FACEMESH_CONTOURS_POINTS = [str(p) for p in
+                                sorted(set([p for p_tup in list(mp_holistic.FACEMESH_CONTOURS) for p in p_tup]))]
+
     dimensions = PoseHeaderDimensions(width=width, height=height, depth=width)
     header = PoseHeader(version=0.1, dimensions=dimensions, components=holistic_components("XYZC", 10))
     body = NumPyPoseBody(fps=10,

sign_language_datasets/datasets/wmt_slt/wmt_slt.py

@@ -15,7 +15,6 @@
     convert_srt_time_to_frame
 
 from ...datasets import SignDatasetConfig
-import srt
 
 _DESCRIPTION = """
 These are Standard German daily news (Tagesschau) and Swiss German weather forecast (Meteo) episodes broadcast and interpreted into Swiss German Sign Language by hearing interpreters (among them, children of Deaf adults, CODA) via Swiss National TV (Schweizerisches Radio und Fernsehen, SRF) (https://www.srf.ch/play/tv/sendung/tagesschau-in-gebaerdensprache?id=c40bed81-b150-0001-2b5a-1e90e100c1c0). For a more extended description of the data, visit https://www.wmt-slt.com/data.
@@ -182,6 +181,14 @@ def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     def _generate_examples(self, datasets):
         """Yields examples."""
 
+        try:
+            import srt
+
+        except ImportError:
+            raise ImportError(
+                "Please install srt with: pip install srt"
+            )
+
         for dataset_id, directories in datasets.items():
             names = [n[:-len('.mp4')] for n in os.listdir(directories['videos'])]
             for name in names: