Lite Classification (#749)

Co-authored-by: Nick <[email protected]>

Author: czaloom

.github/workflows/lite-benchmark-evaluations.yml

@@ -19,20 +19,20 @@ jobs:
       - name: install lite
         run: pip install -e .
         working-directory: ./lite
-      # - name: run classification benchmarks
-      #   run: python benchmark_script.py
-      #   working-directory: ./lite/benchmarks/classification
-      # - name: print classification results
-      #   run: |
-      #     export BENCHMARK_RESULTS=$(python -c "import os;import json;print(json.dumps(json.load(open('results.json', 'r')), indent=4));")
-      #     echo "$BENCHMARK_RESULTS"
-      #   working-directory: ./lite/benchmarks/classification
+      - name: run classification benchmarks
+        run: python benchmark_classification.py
+        working-directory: ./lite/benchmarks/
+      - name: print classification results
+        run: |
+          export BENCHMARK_RESULTS=$(python -c "import os;import json;print(json.dumps(json.load(open('clf_results.json', 'r')), indent=4));")
+          echo "$BENCHMARK_RESULTS"
+        working-directory: ./lite/benchmarks/
       - name: run object detection benchmarks
         run: python benchmark_objdet.py
         working-directory: ./lite/benchmarks/
       - name: print object detection results
         run: |
-          export BENCHMARK_RESULTS=$(python -c "import os;import json;print(json.dumps(json.load(open('manager_results.json', 'r')), indent=4));")
+          export BENCHMARK_RESULTS=$(python -c "import os;import json;print(json.dumps(json.load(open('objdet_results.json', 'r')), indent=4));")
           echo "$BENCHMARK_RESULTS"
         working-directory: ./lite/benchmarks/
       - run: make stop-env

.github/workflows/lite-tests-and-coverage.yml

@@ -19,6 +19,20 @@ jobs:
       - uses: actions/setup-python@v4
         with:
           python-version: "3.10"
+      - name: run classification tests and report coverage
+        run: |
+          pip install -e ".[test]"
+          COVERAGE_FILE=.coverage.classification python -m coverage run --omit "tests/*"  -m pytest -v tests/classification/
+          python -m coverage combine
+          python -m coverage report -m
+          python -m coverage json
+          export TOTAL=$(python -c "import json;print(json.load(open('coverage.json'))['totals']['percent_covered_display'])")
+          echo "total=$TOTAL" >> $GITHUB_ENV
+          if (( $TOTAL < 90 )); then
+            echo "Coverage is below 90%"
+            exit 1
+          fi
+        working-directory: ./lite
       - name: run object detection tests and report coverage
         run: |
           pip install -e ".[test]"

lite/benchmarks/benchmark_classification.py

@@ -0,0 +1,254 @@
+import json
+import os
+from dataclasses import dataclass
+from datetime import datetime
+from pathlib import Path
+from time import time
+
+import requests
+from tqdm import tqdm
+from valor_lite.classification import DataLoader, MetricType
+
+
+def time_it(fn):
+    def wrapper(*args, **kwargs):
+        start = time()
+        results = fn(*args, **kwargs)
+        return (time() - start, results)
+
+    return wrapper
+
+
+def download_data_if_not_exists(
+    file_name: str,
+    file_path: Path,
+    url: str,
+):
+    """Download the data from a public bucket if it doesn't exist locally."""
+
+    if not os.path.exists(file_path):
+        response = requests.get(url, stream=True)
+        if response.status_code == 200:
+            total_size = int(response.headers.get("content-length", 0))
+            with open(file_path, "wb") as f:
+                with tqdm(
+                    total=total_size,
+                    unit="B",
+                    unit_scale=True,
+                    unit_divisor=1024,
+                    desc=file_name,
+                ) as pbar:
+                    for chunk in response.iter_content(chunk_size=1024):
+                        if chunk:
+                            f.write(chunk)
+                            pbar.update(1024)
+        else:
+            raise RuntimeError(response)
+    else:
+        print(f"{file_name} already exists locally.")
+
+
+def write_results_to_file(write_path: Path, results: list[dict]):
+    """Write results to results.json"""
+    current_datetime = datetime.now().strftime("%d/%m/%Y %H:%M:%S")
+    if os.path.isfile(write_path):
+        with open(write_path, "r") as file:
+            file.seek(0)
+            data = json.load(file)
+    else:
+        data = {}
+
+    data[current_datetime] = results
+
+    with open(write_path, "w+") as file:
+        json.dump(data, file, indent=4)
+
+
+@time_it
+def ingest(
+    loader: DataLoader,
+    gt_path: Path,
+    pd_path: Path,
+    limit: int,
+    chunk_size: int,
+):
+    accumulated_time = 0.0
+    with open(gt_path, "r") as gf:
+        with open(pd_path, "r") as pf:
+            count = 0
+            groundtruths = []
+            predictions = []
+            for gline, pline in zip(gf, pf):
+
+                # groundtruth
+                gt_dict = json.loads(gline)
+                groundtruths.append(gt_dict)
+
+                # prediction
+                pd_dict = json.loads(pline)
+                predictions.append(pd_dict)
+
+                count += 1
+                if count >= limit and limit > 0:
+                    break
+                elif len(groundtruths) < chunk_size or chunk_size == -1:
+                    continue
+
+                timer, _ = time_it(loader.add_data_from_valor_dict)(
+                    zip(groundtruths, predictions), True
+                )
+                accumulated_time += timer
+                groundtruths = []
+                predictions = []
+
+            if groundtruths:
+                timer, _ = time_it(loader.add_data_from_valor_dict)(
+                    zip(groundtruths, predictions), True
+                )
+                accumulated_time += timer
+
+    return accumulated_time
+
+
+@dataclass
+class Benchmark:
+    limit: int
+    n_datums: int
+    n_groundtruths: int
+    n_predictions: int
+    n_labels: int
+    chunk_size: int
+    ingestion: float
+    preprocessing: float
+    precomputation: float
+    evaluation: float
+    detailed_evaluation: list[tuple[int, float]]
+
+    def result(self) -> dict:
+        return {
+            "limit": self.limit,
+            "n_datums": self.n_datums,
+            "n_groundtruths": self.n_groundtruths,
+            "n_predictions": self.n_predictions,
+            "n_labels": self.n_labels,
+            "chunk_size": self.chunk_size,
+            "ingestion": {
+                "loading_from_file": f"{round(self.ingestion - self.preprocessing, 2)} seconds",
+                "numpy_conversion": f"{round(self.preprocessing, 2)} seconds",
+                "finalization": f"{round(self.precomputation, 2)} seconds",
+                "total": f"{round(self.ingestion + self.precomputation, 2)} seconds",
+            },
+            "base_evaluation": f"{round(self.evaluation, 2)} seconds",
+            "detailed_evaluation": [
+                {
+                    "n_points": 10,
+                    "n_examples": curve[0],
+                    "computation": f"{round(curve[1], 2)} seconds",
+                }
+                for curve in self.detailed_evaluation
+            ],
+        }
+
+
+def run_benchmarking_analysis(
+    limits_to_test: list[int],
+    results_file: str = "clf_results.json",
+    chunk_size: int = -1,
+    ingestion_timeout=30,
+    evaluation_timeout=30,
+):
+    """Time various function calls and export the results."""
+    current_directory = Path(__file__).parent
+    write_path = current_directory / Path(results_file)
+
+    gt_filename = "gt_classification.jsonl"
+    pd_filename = "pd_classification.jsonl"
+
+    # cache data locally
+    for filename in [gt_filename, pd_filename]:
+        file_path = current_directory / Path(filename)
+        url = f"https://pub-fae71003f78140bdaedf32a7c8d331d2.r2.dev/{filename}"
+        download_data_if_not_exists(
+            file_name=filename, file_path=file_path, url=url
+        )
+
+    # iterate through datum limits
+    results = list()
+    for limit in limits_to_test:
+
+        # === Base Evaluation ===
+        loader = DataLoader()
+
+        # ingest + preprocess
+        (ingest_time, preprocessing_time,) = ingest(
+            loader=loader,
+            gt_path=current_directory / Path(gt_filename),
+            pd_path=current_directory / Path(pd_filename),
+            limit=limit,
+            chunk_size=chunk_size,
+        )  # type: ignore - time_it wrapper
+
+        finalization_time, evaluator = time_it(loader.finalize)()
+
+        if ingest_time > ingestion_timeout and ingestion_timeout != -1:
+            raise TimeoutError(
+                f"Base precomputation timed out with limit of {limit}."
+            )
+
+        # evaluate
+        eval_time, _ = time_it(evaluator.evaluate)()
+        if eval_time > evaluation_timeout and evaluation_timeout != -1:
+            raise TimeoutError(
+                f"Base evaluation timed out with {evaluator.n_datums} datums."
+            )
+
+        detail_no_examples_time, _ = time_it(evaluator.evaluate)(
+            metrics_to_return=[*MetricType.base(), MetricType.ConfusionMatrix],
+        )
+        if (
+            detail_no_examples_time > evaluation_timeout
+            and evaluation_timeout != -1
+        ):
+            raise TimeoutError(
+                f"Base evaluation timed out with {evaluator.n_datums} datums."
+            )
+
+        detail_three_examples_time, _ = time_it(evaluator.evaluate)(
+            metrics_to_return=[*MetricType.base(), MetricType.ConfusionMatrix],
+            number_of_examples=3,
+        )
+        if (
+            detail_three_examples_time > evaluation_timeout
+            and evaluation_timeout != -1
+        ):
+            raise TimeoutError(
+                f"Base evaluation timed out with {evaluator.n_datums} datums."
+            )
+
+        results.append(
+            Benchmark(
+                limit=limit,
+                n_datums=evaluator.n_datums,
+                n_groundtruths=evaluator.n_groundtruths,
+                n_predictions=evaluator.n_predictions,
+                n_labels=evaluator.n_labels,
+                chunk_size=chunk_size,
+                ingestion=ingest_time,
+                preprocessing=preprocessing_time,
+                precomputation=finalization_time,
+                evaluation=eval_time,
+                detailed_evaluation=[
+                    (0, detail_no_examples_time),
+                    (3, detail_three_examples_time),
+                ],
+            ).result()
+        )
+
+    write_results_to_file(write_path=write_path, results=results)
+
+
+if __name__ == "__main__":
+
+    run_benchmarking_analysis(
+        limits_to_test=[5000, 5000, 5000],
+    )

lite/benchmarks/benchmark_objdet.py

@@ -67,7 +67,7 @@ def download_data_if_not_exists(
 
 
 def write_results_to_file(write_path: Path, results: list[dict]):
-    """Write results to manager_results.json"""
+    """Write results to json"""
     current_datetime = datetime.now().strftime("%d/%m/%Y %H:%M:%S")
     if os.path.isfile(write_path):
         with open(write_path, "r") as file:
@@ -178,7 +178,7 @@ def result(self) -> dict:
 def run_benchmarking_analysis(
     limits_to_test: list[int],
     combinations: list[tuple[AnnotationType, AnnotationType]] | None = None,
-    results_file: str = "manager_results.json",
+    results_file: str = "objdet_results.json",
     chunk_size: int = -1,
     compute_pr: bool = True,
     compute_detailed: bool = True,

lite/examples/coco-yolo.ipynb renamed to lite/examples/object-detection.ipynb

@@ -12,9 +12,7 @@
         "\n",
         "In this notebook, we'll walk through a detailed example of how you can use Valor to evaluate object detections made on [the COCO Panoptic dataset](https://cocodataset.org/#home). We'll use Ultralytics' `YOLOv8` model to predict what objects exist in various COCO photographs and compare performance between bounding box and image segmentation results.\n",
         "\n",
-        "For a conceptual introduction to Valor, [check out our project overview](https://striveworks.github.io/valor/). For a higher-level example notebook, [check out our \"Getting Started\" notebook](https://github.com/Striveworks/valor/blob/main/examples/getting_started.ipynb).\n",
-        "\n",
-        "Before using this notebook, please ensure that the Valor service is running on your machine (for start-up instructions, [click here](https://striveworks.github.io/valor/getting_started/)). To connect to a non-local instance of Valor, update `client = Client(\"http://0.0.0.0:8000\")` in the first code block to point to the correct URL."
+        "For a conceptual introduction to Valor, [check out our project overview](https://striveworks.github.io/valor/). For a higher-level example notebook, [check out our \"Getting Started\" notebook](https://github.com/Striveworks/valor/blob/main/examples/getting_started.ipynb)."
       ]
     },
     {
@@ -49,19 +47,6 @@
         "from valor_lite.detection import DataLoader, MetricType"
       ]
     },
-    {
-      "attachments": {},
-      "cell_type": "markdown",
-      "id": "a28f5e66",
-      "metadata": {},
-      "source": [
-        "The modules included in `./integrations` are helper modules that demonstrate how to ingest datasets and model inferences into Valor. The depth of each integration varies depending on the use case. \n",
-        "\n",
-        "The `coco_integration` is designed to download, extract, and upload all in one command as you are starting off with all the the data. \n",
-        "\n",
-        "The `yolo_integration` is much simpler; it is a collection of parser functions that convert YOLO model results into Valor types."
-      ]
-    },
     {
       "cell_type": "code",
       "execution_count": 2,