MNT update notebooks

Author: TomDLT

tutorials/notebooks/movies_3T/06_extract_motion_energy.ipynb

@@ -73,7 +73,7 @@
       },
       "outputs": [],
       "source": [
-        "import numpy as np\nfrom moten.io import imagearray2luminance\n\nfrom voxelwise_tutorials.progress_bar import bar\nfrom voxelwise_tutorials.io import load_hdf5_array\n\n\ndef compute_luminance(run_name, size=(96, 96)):\n\n    stimuli_file = os.path.join(directory, 'stimuli', run_name)\n\n    # get the list of batches in the stimuli file\n    with h5py.File(stimuli_file, 'r') as f:\n        keys = list(f.keys())\n        keys.sort()  # sort the batches\n\n    # compute the luminance on each batch\n    luminance = []\n    for key in bar(keys, title=f'compute_luminance({run_name})'):\n        # load the batch of images\n        images = load_hdf5_array(stimuli_file, key=key)\n\n        # ``imagearray2luminance`` uses uint8 arrays\n        if images.dtype != 'uint8':\n            images = np.int_(np.clip(images, 0, 1) * 255).astype(np.uint8)\n\n        # convert RGB images to a single luminance channel\n        luminance.append(imagearray2luminance(images, size=size))\n\n    return np.concatenate(luminance)\n\n\nluminance_train = np.concatenate(\n    [compute_luminance(f\"train_{ii:02d}.hdf\") for ii in range(12)])\nluminance_test = compute_luminance(\"test.hdf\")"
+        "import numpy as np\nfrom moten.io import imagearray2luminance\n\nfrom voxelwise_tutorials.progress_bar import bar\nfrom voxelwise_tutorials.io import load_hdf5_array\n\n\ndef compute_luminance(run_name, size=(96, 96), batch_size=100):\n\n    stimuli_file = os.path.join(directory, 'stimuli', run_name)\n\n    # get the number of images in the stimuli file\n    with h5py.File(stimuli_file, 'r') as f:\n        n_images = f['stimuli'].shape[0]\n\n    # compute the luminance on each batch\n    luminance = np.zeros((n_images, *size))\n    for start in bar(range(0, n_images, batch_size),\n                     title=f'compute_luminance({run_name})'):\n        # load the batch of images\n        batch = slice(start, start + batch_size)\n        images = load_hdf5_array(stimuli_file, key='stimuli', slice=batch)\n\n        # ``imagearray2luminance`` uses uint8 arrays\n        if images.dtype != 'uint8':\n            images = np.int_(np.clip(images, 0, 1) * 255).astype(np.uint8)\n\n        # convert RGB images to a single luminance channel\n        luminance[batch] = imagearray2luminance(images, size=size)\n\n    return luminance\n\n\nluminance_train = np.concatenate(\n    [compute_luminance(f\"train_{ii:02d}.hdf\") for ii in range(12)])\nluminance_test = compute_luminance(\"test.hdf\")"
       ]
     },
     {
@@ -109,7 +109,7 @@
       },
       "outputs": [],
       "source": [
-        "from voxelwise_tutorials.io import save_hdf5_dataset\n\nfeatures_directory = os.path.join(directory, \"features\")\nif not os.path.exists(features_directory):\n    os.makedirs(features_directory)\n\nsave_hdf5_dataset(\n    os.path.join(features_directory, \"motion_energy_recomputed.hdf\"),\n    dataset=dict(X_train=motion_energy_train, X_test=motion_energy_test))"
+        "from voxelwise_tutorials.io import save_hdf5_dataset\n\nfeatures_directory = os.path.join(directory, \"features\")\nif not os.path.exists(features_directory):\n    os.makedirs(features_directory)\n\nsave_hdf5_dataset(\n    os.path.join(features_directory, \"motion_energy_recomputed.hdf\"),\n    dataset=dict(X_train=motion_energy_train, X_test=motion_energy_test,\n                 run_onsets=np.arange(0, 3600, 300)))"
       ]
     },
     {