notebooks/segment_anything_tutorial.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transforms and materialized views: Segment Anything as example\n",
    "\n",
    "[Segment Anything](https://segment-anything.com/) (SA-1B) is an image segmentation dataset containing 1B masks and 11M images. The source dataset has the following file layout:\n",
    "\n",
    "```\n",
    "sa_000000/sa_1.json\n",
    "         /sa_1.jpg\n",
    "         /sa_2.json\n",
    "         /sa_2.jpg\n",
    "         ...\n",
    "sa_000001/...\n",
    "```\n",
    "\n",
    "where each JSON file contains an image information and its annotations. This example builds a mini pipeline that loads, processes, manages, and feeds data to ML frameworks. We add more complexities to show the features of Space:\n",
    "\n",
    "- JSON data are loaded to a `raw dataset` stored in Parquet. Images are not loaded. This dataset is good for running data mutations and OLAP queries on annotations.\n",
    "- The raw dataset is transformed to a `preprocessed images materialized view` stored in [ArrayRecord](https://github.com/google/array_record). When the raw dataset is modified, refreshing the materialized view (MV) synchronizes changes incrementally. Because annotations are excluded here, the images can be shared by different training tasks, by joining with different annotation tables.\n",
    "- Similarly, the raw dataset is also transformed to an `annotations materialized view`.\n",
    "- Both MVs are keyed by `image_id`, they are joined (low cost JOIN of row addresses instead of real data) to re-construct the `<image, annotation>` pairs. A PyTorch style random access data source is built from the JOIN result, for feeding data to ML frameworks."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 2,
     "metadata": {
      "image/png": {
       "width": 600
      }
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(filename=\"pics/space_segment_anything_example.png\", width=600)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load Raw Data into Space Datasets\n",
    "\n",
    "The first steps load JOSN files into a Space dataset. The equivalent PyArrow schema for the Space dataset/JSON file is:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pyarrow as pa\n",
    "\n",
    "segmentation_schema = pa.struct([\n",
    "  (\"size\", pa.list_(pa.int64())),\n",
    "  (\"counts\", pa.string())\n",
    "])\n",
    "\n",
    "annotation_schema = pa.struct([\n",
    "  (\"area\", pa.int64()),\n",
    "  (\"bbox\", pa.list_(pa.float64())),\n",
    "  (\"crop_box\", pa.list_(pa.float64())),\n",
    "  (\"id\", pa.int64()),\n",
    "  (\"point_coords\", pa.list_(pa.list_(pa.float64()))),\n",
    "  (\"predicted_iou\", pa.float64()),\n",
    "  (\"segmentation\", segmentation_schema),\n",
    "  (\"stability_score\", pa.float64())\n",
    "])\n",
    "\n",
    "image_schema = pa.struct([\n",
    "  (\"image_id\", pa.int64()),\n",
    "  (\"file_name\", pa.string()),\n",
    "  (\"width\", pa.int64()),\n",
    "  (\"height\", pa.int64())\n",
    "])\n",
    "\n",
    "# The JSON file's equivalent schema, used for loading JSON.\n",
    "json_input_schema = pa.schema([\n",
    "  (\"image\", image_schema),\n",
    "  (\"annotations\", pa.list_(annotation_schema))\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following method reads JSON files into Arrow tables, one row per file. Each row represents an image with annotations. It simply uses `pyarrow.json.read_json` to load JSON to Arrow."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import Iterator, List\n",
    "\n",
    "import glob\n",
    "import os\n",
    "from pyarrow import json\n",
    "\n",
    "# The folder storing raw Segment Anything data. Change to your data folder.\n",
    "sa_dir = \"/segment_anything\"\n",
    "\n",
    "# Knobs for making things faster.\n",
    "images_per_shard = 100\n",
    "total_shards = 10\n",
    "\n",
    "# A shard list to read, e.g., [\"sa_000000\", ...]\n",
    "def make_iter(shards: List[str]) -> Iterator[pa.Table]:\n",
    "\n",
    "  def json_files_to_arrow(shard: str) -> pa.Table:\n",
    "    batch: List[pa.Table] = []\n",
    "\n",
    "    pattern = os.path.join(sa_dir, shard, \"*.json\")\n",
    "    print(f\"processing pattern: {pattern}\")\n",
    "    for f in glob.glob(pattern):\n",
    "      batch.append(json.read_json(f,\n",
    "        parse_options=json.ParseOptions(explicit_schema=json_input_schema)))\n",
    "      if len(batch) >= images_per_shard:\n",
    "        break\n",
    "\n",
    "    table = pa.concat_tables(batch)\n",
    "    image_column = table.column(\"image\").combine_chunks()\n",
    "    image_id_column = image_column.field(\"image_id\")\n",
    "    file_name_column = image_column.field(\"file_name\")\n",
    "\n",
    "    file_paths = [\n",
    "      os.path.join(shard, file_name.as_py()) for file_name in file_name_column]\n",
    "    file_path_column = pa.StringArray.from_pandas(file_paths)\n",
    "\n",
    "    # Add image_id and file_path columns to the original data.\n",
    "    table = table.append_column(pa.field(\"file_path\", pa.string()), file_path_column)\n",
    "    table = table.append_column(pa.field(\"image_id\", pa.int64()), image_id_column)\n",
    "    return table\n",
    "\n",
    "  for shard in shards:\n",
    "    yield json_files_to_arrow(shard)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create an empty Space dataset:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from space import Dataset\n",
    "# A Space column field type representing a file.\n",
    "from space import File\n",
    "\n",
    "# This is the top level dataset schema.\n",
    "ds_schema = pa.schema([\n",
    "  (\"image_id\", pa.int64()),  # Add an image_id as the primary key.\n",
    "  (\"file_path\", File(directory=sa_dir)),  # A field representing an external file.\n",
    "  (\"image\", image_schema),\n",
    "  (\"annotations\", pa.list_(annotation_schema))\n",
    "])\n",
    "\n",
    "raw_ds_location = \"/space/datasets/raw_data\"\n",
    "\n",
    "# `record_fields` specifies which fields to store in ArrayRecord.\n",
    "# The raw dataset is pure Parquet, so leave it empty.\n",
    "raw_ds = Dataset.create(raw_ds_location, ds_schema,\n",
    "  primary_keys=[\"image_id\"], record_fields=[])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After creation, the dataset can be loaded from location later:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "raw_ds = Dataset.load(raw_ds_location)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now start loading JSON data into Space distributedly using Ray:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# append_from accepts a list of no-args methods that returns a generator.\n",
    "# A method is used because generator cannot be pickled for distributed execuion.\n",
    "raw_ds.ray().append_from(\n",
    "  [lambda idx=i: make_iter([f\"sa_{idx:06}\"]) for i in range(total_shards)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pre-processed Image MVs\n",
    "\n",
    "After loading image information and annotations into Space, next step we read and pre-process images, then persist the result in a Space Materialized View (MV).\n",
    "\n",
    "First define a function to load and process images for a batch of inputs. The Ray runner uses [Ray map_batches](https://docs.ray.io/en/latest/data/api/doc/ray.data.Dataset.map_batches.html) internally, so the input format is `Dict[str, np.ndarray]`, where key is field names in the raw dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import Any, Dict\n",
    "import cv2\n",
    "\n",
    "# The type records directory information.\n",
    "# In future we plan to add more file features to auto read/decode data.\n",
    "file_type = raw_ds.schema.field(\"file_path\").type\n",
    "\n",
    "def read_and_preprocess_image(data: Dict[str, Any]) -> Dict[str, Any]:\n",
    "  ims = []\n",
    "  for file_path in data[\"file_path\"]:\n",
    "    im = cv2.imread(file_type.full_path(file_path))\n",
    "    im = cv2.resize(im, dsize=(100, 100), interpolation=cv2.INTER_CUBIC)\n",
    "    ims.append(im.tobytes())\n",
    "\n",
    "  # Drop `file_path` column that won't write into the output view, add a new\n",
    "  # `image_bytes` column.\n",
    "  del data[\"file_path\"]\n",
    "  data[\"image_bytes\"] = ims\n",
    "  return data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a view `image_view` by transforming the raw dataset using the function above. By providing an `input_fields`, we only read these fields from the dataset to save IO (annotations are not read). The output schema must be compatible with the data returned by `read_and_preprocess_image`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a view by transforming the source dataset.\n",
    "image_view = raw_ds.map_batches(\n",
    "  fn=read_and_preprocess_image,\n",
    "  input_fields=[\"image_id\", \"file_path\"], # Exclude the `annotations`, `image` fields\n",
    "  output_schema=pa.schema([\n",
    "    (\"image_id\", pa.int64()),\n",
    "    (\"image_bytes\", pa.binary())  # Add a new field for image bytes\n",
    "  ]),\n",
    "  output_record_fields=[\"image_bytes\"]  # Store this field in ArrayRecord\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Queries on `image_view` reads all data from `raw_ds` and running the transform again. To avoid that, we can **materialize** the view:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create an empty MV.\n",
    "image_mv_location = \"/space/mvs/images\"\n",
    "image_mv = image_view.materialize(image_mv_location)\n",
    "\n",
    "# Synchronize the MV to the source dataset. It populates the MV storage.\n",
    "image_mv.ray().refresh()\n",
    "\n",
    "# Verifly the images in MV.\n",
    "image_mv.ray().read_all().num_rows"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Same as datasets, MVs can be loaded from file location anytime:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from space import MaterializedView\n",
    "\n",
    "image_mv = MaterializedView.load(image_mv_location)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Make some modifications in the source dataset, and synchronize changes to MV."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Show all image IDs, and pick a few to delete.\n",
    "raw_ds.local().read_all().select([\"image_id\"])\n",
    "\n",
    "# Delete two images.\n",
    "raw_ds.local().delete((pc.field(\"image_id\") == 4811) | (pc.field(\"image_id\") == 6973))\n",
    "\n",
    "image_mv.ray().refresh()\n",
    "# The images are deleted in MV.\n",
    "image_mv.local().read_all().select([\"image_id\"]).num_rows"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Annotations MVs\n",
    "\n",
    "This step creates a MV for annotations. Because there can be hundreds of objects with segmentation data in annotations, its size is much larger than other datasets. This example stores serialized annotations (`annotations_bytes`) in ArrayRecord."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "def serialize_annotations(data: Dict[str, Any]) -> Dict[str, Any]:\n",
    "  # pickle is just an example of serialization.\n",
    "  data[\"annotations_bytes\"] = [pickle.dumps(d) for d in data[\"annotations\"]]\n",
    "  return data\n",
    "\n",
    "\n",
    "# Create a view by transforming the source dataset.\n",
    "annotations_view = ds.map_batches(\n",
    "  fn=serialize_annotations,\n",
    "  input_fields=[\"image_id\", \"annotations\"],\n",
    "  output_schema=pa.schema([\n",
    "    (\"image_id\", pa.int64()),\n",
    "    (\"annotations_bytes\", pa.binary())\n",
    "  ]),\n",
    "  output_record_fields=[\"annotations_bytes\"]\n",
    ")\n",
    "\n",
    "# Create an empty MV.\n",
    "annotations_mv_location = \"/space/mvs/annotations\"\n",
    "annotations_mv = annotations_view.materialize(annotations_mv_location)\n",
    "\n",
    "annotations_mv.ray().refresh()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Joining Images and Annotations data\n",
    "\n",
    "Images and annotations are now stored separately. At training time, we have to pair they together and feed them to the training framework. Both images and annotations MVs have primary key `image_id`, so let's join them.\n",
    "\n",
    "JOIN is relatively cheap in Space, because it can:\n",
    "\n",
    "- Read row addresses of records in ArrayRecord files instead of data.\n",
    "- Joining addresses.\n",
    "- Build a random access data source on top of the JOIN result, which auto reads data from the addresses.\n",
    "\n",
    "When the data to join does not fit into memory, we can use a **partitioned** JOIN to run it distributedly by providing a `partition_fn`. `partition_fn` is used to divide the whole range of join key into smaller ranges to run on multiple workers. In future we may provide auto range partitioning based on storage statistics."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from space import JoinOptions, Range\n",
    "\n",
    "# left_reference_read, right_reference_read enables reference read.\n",
    "# Addresses are read for the join inputs, instead of reading data.\n",
    "joined_view = image_mv.dataset.join(annotations_mv.dataset, keys=[\"image_id\"],\n",
    "  left_reference_read=True, right_reference_read=True)\n",
    "\n",
    "def partition_fn(r: Range):\n",
    "  mid = (r.min_ + r.max_) // 2\n",
    "  return [Range(r.min_, mid, include_max=False), Range(mid, r.max_, include_max=True)]\n",
    "\n",
    "joined_addresses = joined_view.ray().read_all(\n",
    "  join_options=JoinOptions(partition_fn=partition_fn))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The JOIN result table contains record addresses of `image_bytes` and `annotations_bytes`. Space provides a PyTorch style random access data source for reading the addresses:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from space import RandomAccessDataSource\n",
    "\n",
    "datasource = RandomAccessDataSource(\n",
    "  # field-name: storage-location, for reading data from ArrayRecord files.\n",
    "  {\n",
    "    \"image_bytes\": image_mv_location,\n",
    "    \"annotations_bytes\": annotations_mv_location,\n",
    "  },\n",
    "  joined_addresses,\n",
    "  # Don't auto deserialize data, because we store them as plain bytes.\n",
    "  deserialize=False)\n",
    "\n",
    "# Total data size.\n",
    "len(datasource)\n",
    "\n",
    "# Random access records.\n",
    "datasource[89].keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The example stores annotations and images as bytes. Space has build-in support for [TFDS FeaturesDict](https://www.tensorflow.org/datasets/api_docs/python/tfds/features/FeaturesDict) to make serializing/deserializing data more managed, see the [TFDS example](tfds_coco_tutorial.ipynb). "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}