Formatting and Style (#53)

* updated pre-commit hooks * initial formatting * fix typo * style changes * reverted
luxonis · Dec 19, 2023 · 4e8ca1c · 4e8ca1c
1 parent 40f4b36
commit 4e8ca1c
Show file tree

Hide file tree

Showing 51 changed files with 890 additions and 2,399 deletions.
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -1,11 +1,30 @@
 repos:
-  - repo: https://github.com/ambv/black
-    rev: 23.3.0
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.1.2
     hooks:
-    - id: black
-      language_version: python3.8
+      - id: ruff
+        args: [--fix, --exit-non-zero-on-fix]
+        types_or: [python, pyi, jupyter]
+      - id: ruff-format
+        types_or: [python, pyi, jupyter]
+
+  - repo: https://github.com/PyCQA/docformatter
+    rev: v1.7.5
+    hooks:
+      - id: docformatter
+        additional_dependencies: [tomli]
+        args: [--in-place]
+
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.4.0
     hooks:
     - id: no-commit-to-branch
       args: ['--branch', 'main', '--branch', 'dev']
+
+  - repo: https://github.com/executablebooks/mdformat
+    rev: 0.7.10
+    hooks:
+      - id: mdformat
+        additional_dependencies:
+          - mdformat-gfm
+          - mdformat-toc
diff --git a/README.md b/README.md
@@ -104,8 +104,8 @@ You could see that this configuration could also be used for any OAK-D-like prod
 
 Some more definitions:
 
-* `HType`\: The type of the component. Right now, there's really only `IMAGE`. But eventually, we will have video and point cloud.
-* `IType`\: For an image component, the type of the image. `BGR` is for any 3-channel image, `MONO` is for single-channel images, and `DEPTH` and `DISPARITY` can be uint16 single-channel images for subpixel disparity. The only difference between `DISPARITY` and `DEPTH` is a semantic difference.
+- `HType`: The type of the component. Right now, there's really only `IMAGE`. But eventually, we will have video and point cloud.
+- `IType`: For an image component, the type of the image. `BGR` is for any 3-channel image, `MONO` is for single-channel images, and `DEPTH` and `DISPARITY` can be uint16 single-channel images for subpixel disparity. The only difference between `DISPARITY` and `DEPTH` is a semantic difference.
 
 #### Adding Data
 
@@ -177,8 +177,8 @@ with LuxonisDataset(team_name, dataset_name) as dataset:
 
 The first element in the loaded tuple is the image(s) and the second is a dictionary where the keys may provide annotations in different formats.
 
-* `imgs` is (batch size x number of channels x height x width)
-* `dict['class']` is (batch size x number of classes)
-* `dict['bbox']` is (number of boxes in batch x 6 [image ID, class, x, y, width, height])
-* `dict['segmentation']` is (batch size x number of classes x height x width)
-* `dict['keypoints']` is (batch size x (number of points*2 + 1))
+- `imgs` is (batch size x number of channels x height x width)
+- `dict['class']` is (batch size x number of classes)
+- `dict['bbox']` is (number of boxes in batch x 6 \[image ID, class, x, y, width, height\])
+- `dict['segmentation']` is (batch size x number of classes x height x width)
+- `dict['keypoints']` is (batch size x (number of points\*2 + 1))
diff --git a/examples/COCO_people_subset.ipynb b/examples/COCO_people_subset.ipynb
@@ -15,15 +15,17 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "import glob, json, os\n",
+    "import glob\n",
+    "import json\n",
     "import numpy as np\n",
     "import cv2\n",
-    "from copy import deepcopy\n",
     "from tqdm import tqdm\n",
     "import matplotlib.pyplot as plt\n",
-    "from pycocotools import mask as mask_utils\n",
+    "import os\n",
+    "import zipfile\n",
+    "import gdown\n",
     "\n",
-    "from luxonis_ml.data import *\n",
+    "from luxonis_ml.data import LuxonisDataset, LuxonisLoader\n",
     "from luxonis_ml.enums import LabelType"
    ]
   },
@@ -59,9 +61,22 @@
    },
    "outputs": [],
    "source": [
-    "! pip install gdown\n",
-    "! gdown 1XlvFK7aRmt8op6-hHkWVKIJQeDtOwoRT -O ../data/COCO_people_subset.zip\n",
-    "! unzip ../data/COCO_people_subset.zip -d ../data/"
+    "url = \"https://drive.google.com/uc?id=1XlvFK7aRmt8op6-hHkWVKIJQeDtOwoRT\"\n",
+    "output_zip = \"../data/COCO_people_subset.zip\"\n",
+    "output_folder = \"../data/\"\n",
+    "\n",
+    "# Check if the data already exists\n",
+    "if not os.path.exists(output_zip) and not os.path.exists(\n",
+    "    os.path.join(output_folder, \"COCO_people_subset\")\n",
+    "):\n",
+    "    # Download the file\n",
+    "    gdown.download(url, output_zip, quiet=False)\n",
+    "\n",
+    "    # Unzip the file\n",
+    "    with zipfile.ZipFile(output_zip, \"r\") as zip_ref:\n",
+    "        zip_ref.extractall(output_folder)\n",
+    "else:\n",
+    "    print(\"Data already exists. Exiting.\")"
    ]
   },
   {
@@ -104,88 +119,86 @@
     "# splits = ['train' for _ in range(20)] + ['val' for _ in range(10)]\n",
     "\n",
     "def COCO_people_subset_generator():\n",
-    "\n",
     "    # find image paths and load COCO annotations\n",
-    "    img_dir = '../data/person_val2017_subset'\n",
-    "    annot_file = '../data/person_keypoints_val2017.json'\n",
+    "    img_dir = \"../data/person_val2017_subset\"\n",
+    "    annot_file = \"../data/person_keypoints_val2017.json\"\n",
     "    # get paths to images sorted by number\n",
-    "    im_paths = glob.glob(img_dir+'/*.jpg')\n",
-    "    nums = np.array([int(path.split('/')[-1].split('.')[0]) for path in im_paths])\n",
+    "    im_paths = glob.glob(img_dir + \"/*.jpg\")\n",
+    "    nums = np.array([int(path.split(\"/\")[-1].split(\".\")[0]) for path in im_paths])\n",
     "    idxs = np.argsort(nums)\n",
     "    im_paths = list(np.array(im_paths)[idxs])\n",
-    "    # load \n",
+    "    # load\n",
     "    with open(annot_file) as file:\n",
     "        data = json.load(file)\n",
-    "    imgs = data['images']\n",
-    "    anns = data['annotations']\n",
-    "    \n",
+    "    imgs = data[\"images\"]\n",
+    "    anns = data[\"annotations\"]\n",
+    "\n",
     "    for i, path in tqdm(enumerate(im_paths)):\n",
     "        # find annotations matching the COCO image\n",
-    "        gran = path.split('/')[-1]\n",
-    "        img = [img for img in imgs if img['file_name']==gran][0]\n",
-    "        img_id = img['id']\n",
-    "        img_anns = [ann for ann in anns if ann['image_id'] == img_id]\n",
-    "        \n",
+    "        gran = path.split(\"/\")[-1]\n",
+    "        img = [img for img in imgs if img[\"file_name\"] == gran][0]\n",
+    "        img_id = img[\"id\"]\n",
+    "        img_anns = [ann for ann in anns if ann[\"image_id\"] == img_id]\n",
+    "\n",
     "        # load the image\n",
     "        im = cv2.imread(path)\n",
     "        height, width, _ = im.shape\n",
-    "        \n",
-    "        # initialize annotations for LDF\n",
-    "        mask = np.zeros((height, width)) # segmentation mask is always a HxW numpy array\n",
-    "        boxes = [] # bounding boxes are a list of [class, x, y, width, height] of the box\n",
-    "        keypoints = [] # keypoints are a list of classes and (x,y) points\n",
-    "    \n",
+    "\n",
     "        if len(img_anns):\n",
     "            yield {\n",
     "                \"file\": path,\n",
     "                \"class\": \"person\",\n",
     "                \"type\": \"classification\",\n",
-    "                \"value\": True\n",
+    "                \"value\": True,\n",
     "            }\n",
-    "        \n",
+    "\n",
     "        for ann in img_anns:\n",
     "            # COCO-specific conversion for segmentation\n",
-    "            seg = ann['segmentation']\n",
-    "            if isinstance(seg, list): # polyline format\n",
+    "            seg = ann[\"segmentation\"]\n",
+    "            if isinstance(seg, list):  # polyline format\n",
     "                poly = []\n",
     "                for s in seg:\n",
-    "                    poly_arr = np.array(s).reshape(-1,2)\n",
-    "                    poly += [(poly_arr[i,0]/width, poly_arr[i,1]/height) for i in range(len(poly_arr))]\n",
+    "                    poly_arr = np.array(s).reshape(-1, 2)\n",
+    "                    poly += [\n",
+    "                        (poly_arr[i, 0] / width, poly_arr[i, 1] / height)\n",
+    "                        for i in range(len(poly_arr))\n",
+    "                    ]\n",
     "                yield {\n",
     "                    \"file\": path,\n",
     "                    \"class\": \"person\",\n",
     "                    \"type\": \"polyline\",\n",
-    "                    \"value\": poly\n",
+    "                    \"value\": poly,\n",
     "                }\n",
-    "            else: # RLE format\n",
+    "            else:  # RLE format\n",
     "                value = (seg[\"size\"][0], seg[\"size\"][1], seg[\"counts\"])\n",
     "                yield {\n",
     "                    \"file\": path,\n",
     "                    \"class\": \"person\",\n",
     "                    \"type\": \"segmentation\",\n",
-    "                    \"value\": value\n",
+    "                    \"value\": value,\n",
     "                }\n",
-    "                \n",
-    "            \n",
+    "\n",
     "            # COCO-specific conversion for bounding boxes\n",
-    "            x, y, w, h = ann['bbox']\n",
+    "            x, y, w, h = ann[\"bbox\"]\n",
     "            yield {\n",
     "                \"file\": path,\n",
     "                \"class\": \"person\",\n",
     "                \"type\": \"box\",\n",
-    "                \"value\": (x/width, y/height, w/width, h/height)\n",
+    "                \"value\": (x / width, y / height, w / width, h / height),\n",
     "            }\n",
-    "            \n",
+    "\n",
     "            # COCO-specific conversion for keypoints\n",
-    "            kps = np.array(ann['keypoints']).reshape(-1, 3)\n",
+    "            kps = np.array(ann[\"keypoints\"]).reshape(-1, 3)\n",
     "            keypoint = []\n",
     "            for kp in kps:\n",
-    "                keypoint.append((float(kp[0]/width), float(kp[1]/height), int(kp[2])))  \n",
+    "                keypoint.append(\n",
+    "                    (float(kp[0] / width), float(kp[1] / height), int(kp[2]))\n",
+    "                )\n",
     "            yield {\n",
     "                \"file\": path,\n",
     "                \"class\": \"person\",\n",
     "                \"type\": \"keypoints\",\n",
-    "                \"value\": keypoint\n",
+    "                \"value\": keypoint,\n",
     "            }"
    ]
   },
@@ -199,15 +212,17 @@
     "dataset = LuxonisDataset(dataset_name)\n",
     "dataset.set_classes([\"person\"])\n",
     "\n",
-    "annot_file = '../data/person_keypoints_val2017.json'\n",
+    "annot_file = \"../data/person_keypoints_val2017.json\"\n",
     "with open(annot_file) as file:\n",
     "    data = json.load(file)\n",
-    "dataset.set_skeletons({\n",
-    "    \"person\": {\n",
-    "        \"labels\": data[\"categories\"][0]['keypoints'],\n",
-    "        \"edges\": (np.array(data[\"categories\"][0][\"skeleton\"])-1).tolist()\n",
+    "dataset.set_skeletons(\n",
+    "    {\n",
+    "        \"person\": {\n",
+    "            \"labels\": data[\"categories\"][0][\"keypoints\"],\n",
+    "            \"edges\": (np.array(data[\"categories\"][0][\"skeleton\"]) - 1).tolist(),\n",
+    "        }\n",
     "    }\n",
-    "})\n",
+    ")\n",
     "dataset.add(COCO_people_subset_generator)"
    ]
   },
@@ -251,7 +266,7 @@
     "    box = ann[LabelType.BOUNDINGBOX]\n",
     "    seg = ann[LabelType.SEGMENTATION]\n",
     "    kps = ann[LabelType.KEYPOINT]\n",
-    "    \n",
+    "\n",
     "    # print(\"Sample classification tensor\")\n",
     "    # print(cls)\n",
     "    # print()\n",
@@ -270,19 +285,25 @@
     "\n",
     "    h, w, _ = image.shape\n",
     "    for b in box:\n",
-    "        cv2.rectangle(image, (int(b[1]*w),int(b[2]*h)), (int(b[1]*w+b[3]*w),int(b[2]*h+b[4]*h)), (255,0,0), 2)\n",
-    "    mask_viz = np.zeros((h,w,3)).astype(np.uint8)\n",
+    "        cv2.rectangle(\n",
+    "            image,\n",
+    "            (int(b[1] * w), int(b[2] * h)),\n",
+    "            (int(b[1] * w + b[3] * w), int(b[2] * h + b[4] * h)),\n",
+    "            (255, 0, 0),\n",
+    "            2,\n",
+    "        )\n",
+    "    mask_viz = np.zeros((h, w, 3)).astype(np.uint8)\n",
     "    for mask in seg:\n",
-    "        mask_viz[mask==1, 2] = 255\n",
+    "        mask_viz[mask == 1, 2] = 255\n",
     "    image = cv2.addWeighted(image, 0.5, mask_viz, 0.5, 0)\n",
     "\n",
     "    for kp in kps:\n",
-    "        kp = kp[1:].reshape(-1,3)\n",
+    "        kp = kp[1:].reshape(-1, 3)\n",
     "        for k in kp:\n",
-    "            cv2.circle(image, (int(k[0]*w),int(k[1]*h)), 2, (0,255,0), 2)\n",
-    "    \n",
+    "            cv2.circle(image, (int(k[0] * w), int(k[1] * h)), 2, (0, 255, 0), 2)\n",
+    "\n",
     "    plt.imshow(image)\n",
-    "    plt.axis('off')  # Optional: Hide axis\n",
+    "    plt.axis(\"off\")  # Optional: Hide axis\n",
     "    plt.show()\n",
     "    # break"
    ]