labelr 0.10.0__py3-none-any.whl → 0.11.1__py3-none-any.whl

labelr/apps/label_studio.py

@@ -7,25 +7,39 @@ from typing import Annotated, Optional
 import typer
 from openfoodfacts.utils import get_logger
 
-from ..config import LABEL_STUDIO_DEFAULT_URL
+from . import typer_description
+from ..config import config
 
 app = typer.Typer()
 
 logger = get_logger(__name__)
 
 
+def check_label_studio_api_key(api_key: str | None):
+    if not api_key:
+        raise typer.BadParameter(
+            "Label Studio API key not provided. Please provide it with the "
+            "--api-key option or set the LABELR_LABEL_STUDIO_API_KEY environment variable."
+        )
+
+
 @app.command()
 def create(
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
     title: Annotated[str, typer.Option(help="Project title")],
     config_file: Annotated[
         Path, typer.Option(help="Path to label config file", file_okay=True)
     ],
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
 ):
     """Create a new Label Studio project."""
     from label_studio_sdk.client import LabelStudio
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
     label_config = config_file.read_text()
 
@@ -35,7 +49,6 @@ def create(
 
 @app.command()
 def import_data(
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
     project_id: Annotated[int, typer.Option(help="Label Studio Project ID")],
     dataset_path: Annotated[
         Path,
@@ -43,8 +56,15 @@ def import_data(
             help="Path to the Label Studio dataset JSONL file", file_okay=True
         ),
     ],
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
-    batch_size: int = 25,
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
+    batch_size: Annotated[
+        int, typer.Option(help="Number of tasks to import as a single batch")
+    ] = 25,
 ):
     """Import tasks from a dataset file to a Label Studio project.
 
@@ -56,6 +76,7 @@ def import_data(
     import tqdm
     from label_studio_sdk.client import LabelStudio
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
 
     with dataset_path.open("rt") as f:
@@ -67,12 +88,17 @@ def import_data(
 
 @app.command()
 def update_prediction(
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
     project_id: Annotated[int, typer.Option(help="Label Studio project ID")],
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
 ):
     from label_studio_sdk.client import LabelStudio
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
 
     for task in ls.tasks.list(project=project_id, fields="all"):
@@ -91,16 +117,7 @@ def add_split(
     train_split: Annotated[
         float, typer.Option(help="fraction of samples to add in train split")
     ],
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
     project_id: Annotated[int, typer.Option(help="Label Studio project ID")],
-    split_name: Annotated[
-        Optional[str],
-        typer.Option(
-            help="name of the split associated "
-            "with the task ID file. If --task-id-file is not provided, "
-            "this field is ignored."
-        ),
-    ] = None,
     train_split_name: Annotated[
         str,
         typer.Option(help="name of the train split"),
@@ -110,13 +127,33 @@ def add_split(
         typer.Option(help="name of the validation split"),
     ] = "val",
     task_id_file: Annotated[
-        Optional[Path],
+        Path | None,
         typer.Option(help="path of a text file containing IDs of samples"),
     ] = None,
+    split_name: Annotated[
+        str | None,
+        typer.Option(
+            help="name of the split associated "
+            "with the task ID file. If --task-id-file is not provided, "
+            "this field is ignored."
+        ),
+    ] = None,
     overwrite: Annotated[
         bool, typer.Option(help="overwrite existing split field")
     ] = False,
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    view_id: Annotated[
+        int | None,
+        typer.Option(
+            help="ID of the Label Studio view, if any. This option is useful "
+            "to filter the task to process."
+        ),
+    ] = None,
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
 ):
     """Update the split field of tasks in a Label Studio project.
 
@@ -124,7 +161,7 @@ def add_split(
 
     If `--task-id-file` is provided, it should contain a list of task IDs,
     one per line. The split field of these tasks will be updated to the value
-    of `--split-name`.
+    of `--split-name`. The `--train-split` value is ignored in this case.
 
     If `--task-id-file` is not provided, the split field of all tasks in the
     project will be updated based on the `train_split` probability.
@@ -133,12 +170,16 @@ def add_split(
 
     In both cases, tasks with a non-null split field are not updated unless
     the `--overwrite` flag is provided.
+
+    The `--view-id` option can be used to only assign the split on a subset
+    of the tasks.
     """
     import random
 
     from label_studio_sdk import Task
     from label_studio_sdk.client import LabelStudio
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
 
     task_ids = None
@@ -149,14 +190,17 @@ def add_split(
             )
         task_ids = task_id_file.read_text().strip().split("\n")
 
-    for task in ls.tasks.list(project=project_id, fields="all"):
+    for task in ls.tasks.list(project=project_id, fields="all", view=view_id):
         task: Task
         task_id = task.id
 
         split = task.data.get("split")
         if split is None or overwrite:
-            if task_ids and str(task_id) in task_ids:
-                split = split_name
+            if task_ids:
+                if str(task_id) in task_ids:
+                    split = split_name
+                else:
+                    continue
             else:
                 split = (
                     train_split_name
@@ -170,12 +214,16 @@ def add_split(
 
 @app.command()
 def annotate_from_prediction(
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
     project_id: Annotated[int, typer.Option(help="Label Studio project ID")],
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
     updated_by: Annotated[
         Optional[int], typer.Option(help="User ID to declare as annotator")
     ] = None,
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
 ):
     """Create annotations for all tasks from predictions.
 
@@ -186,6 +234,7 @@ def annotate_from_prediction(
     from label_studio_sdk.client import LabelStudio
     from label_studio_sdk.types.task import Task
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
 
     task: Task
@@ -203,42 +252,66 @@ def annotate_from_prediction(
             )
 
 
-class PredictorBackend(enum.Enum):
-    ultralytics = "ultralytics"
-    robotoff = "robotoff"
+class PredictorBackend(enum.StrEnum):
+    ultralytics = enum.auto()
+    ultralytics_yolo_world = enum.auto()
+    ultralytics_sam3 = enum.auto()
+
+
+YOLO_WORLD_MODELS = (
+    "yolov8s-world.pt",
+    "yolov8s-worldv2.pt",
+    "yolov8m-world.pt",
+    "yolov8m-worldv2.pt",
+    "yolov8l-world.pt",
+    "yolov8l-worldv2.pt",
+    "yolov8x-world.pt",
+    "yolov8x-worldv2.pt",
+)
 
 
 @app.command()
 def add_prediction(
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
     project_id: Annotated[int, typer.Option(help="Label Studio Project ID")],
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
     view_id: Annotated[
-        Optional[int],
+        int | None,
         typer.Option(
             help="Label Studio View ID to filter tasks. If not provided, all tasks in the "
             "project are processed."
         ),
     ] = None,
     model_name: Annotated[
-        str,
+        str | None,
         typer.Option(
-            help="Name of the object detection model to run (for Robotoff server) or "
-            "of the Ultralytics zero-shot model to run."
+            "--model",
+            help="Name or path of the object detection model to run. How this is used depends "
+            "on the backend. If using `ultralytics` backend, the option is required and is the "
+            "name of the model to download from the Ultralytics model zoo or the path to a local "
+            "model. "
+            "If using `ultralytics_yolo_world` backend, this is optional and is the name of the "
+            "`yolo-world` model to download from the Ultralytics model zoo or the path to a local "
+            "model (Defaults: `yolov8x-worldv2.pt`). "
+            "If using `ultralytics_sam3` backend, this option is ignored, as there is a single model. "
+            "The model is downloaded automatically from Hugging Face.",
         ),
-    ] = "yolov8x-worldv2.pt",
-    server_url: Annotated[
-        Optional[str],
-        typer.Option(help="The Robotoff URL if the backend is robotoff"),
-    ] = "https://robotoff.openfoodfacts.org",
+    ] = None,
+    skip_existing: Annotated[
+        bool,
+        typer.Option(
+            help="Skip tasks that already have predictions",
+        ),
+    ] = True,
     backend: Annotated[
         PredictorBackend,
         typer.Option(
-            help="Prediction backend: either use Ultralytics to perform "
-            "the prediction or Robotoff server."
+            help="The prediction backend, possible options are: `ultralytics` or `ultralytics_sam3`"
         ),
     ] = PredictorBackend.ultralytics,
     labels: Annotated[
-        Optional[list[str]],
+        list[str] | None,
         typer.Option(
             help="List of class labels to use for Yolo model. If you're using Yolo-World or other "
             "zero-shot models, this is the list of label names that are going to be provided to the "
@@ -247,15 +320,19 @@ def add_prediction(
         ),
     ] = None,
     label_mapping: Annotated[
-        Optional[str],
-        typer.Option(help="Mapping of model labels to class names, as a JSON string"),
+        str | None,
+        typer.Option(
+            help='Mapping of model labels to class names, as a JSON string. Example: \'{"price tag": "price-tag"}\''
+        ),
     ] = None,
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
     threshold: Annotated[
-        Optional[float],
+        float | None,
         typer.Option(
-            help="Confidence threshold for selecting bounding boxes. The default is 0.3 "
-            "for robotoff backend and 0.1 for ultralytics backend."
+            help="Confidence threshold for selecting bounding boxes. The default is 0.1 for "
+            "ultralytics backend and 0.25 for ultralytics_sam3 backend."
         ),
     ] = None,
     max_det: Annotated[int, typer.Option(help="Maximum numbers of detections")] = 300,
@@ -270,22 +347,31 @@ def add_prediction(
         typer.Option(help="Raise an error if image download fails"),
     ] = True,
     model_version: Annotated[
-        Optional[str],
-        typer.Option(help="Model version to use for the prediction"),
+        str | None,
+        typer.Option(
+            help="Set the model version field of the prediction sent to Label Studio. "
+            "This is used to track which model generated the prediction."
+        ),
+    ] = None,
+    imgsz: Annotated[
+        int | None,
+        typer.Option(
+            help="Image size to use for Ultralytics models. If not provided, "
+            "the default size of the model is used."
+        ),
     ] = None,
 ):
-    """Add predictions as pre-annotations to Label Studio tasks,
-    for an object detection model running on Triton Inference Server."""
+    """Add predictions as pre-annotations to Label Studio tasks."""
 
     import tqdm
+    from huggingface_hub import hf_hub_download
     from label_studio_sdk.client import LabelStudio
-    from openfoodfacts.utils import get_image_from_url, http_session
+    from openfoodfacts.utils import get_image_from_url
     from PIL import Image
 
-    from ..annotate import (
-        format_annotation_results_from_robotoff,
-        format_annotation_results_from_ultralytics,
-    )
+    from ..annotate import format_annotation_results_from_ultralytics
+
+    check_label_studio_api_key(api_key)
 
     label_mapping_dict = None
     if label_mapping:
@@ -294,6 +380,13 @@ def add_prediction(
     if dry_run:
         logger.info("** Dry run mode enabled **")
 
+    if backend == PredictorBackend.ultralytics and not Path(model_name).is_file():
+        raise typer.BadParameter(
+            f"Model file '{model_name}' not found. When the backend is `ultralytics` "
+            "and the --model does not refer to a YOLO-World model, --model is expected "
+            "to be a local Ultralytics model file (`.pt`)."
+        )
+
     logger.info(
         "backend: %s, model_name: %s, labels: %s, threshold: %s, label mapping: %s",
         backend,
@@ -304,66 +397,83 @@ def add_prediction(
     )
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
 
-    if backend == PredictorBackend.ultralytics:
-        from ultralytics import YOLO
+    if backend in (
+        PredictorBackend.ultralytics,
+        PredictorBackend.ultralytics_yolo_world,
+    ):
+        from ultralytics import YOLO, YOLOWorld
 
         if labels is None:
-            raise typer.BadParameter("Labels are required for Ultralytics backend")
+            raise typer.BadParameter("Labels are required for `ultralytics` backend")
 
         if threshold is None:
             threshold = 0.1
 
-        model = YOLO(model_name)
-        if hasattr(model, "set_classes"):
+        if backend == PredictorBackend.ultralytics:
+            model = YOLO(model_name)
+        elif backend == PredictorBackend.ultralytics_yolo_world:
+            if model_name is None:
+                model_name = "yolov8x-worldv2.pt"
+            model = YOLOWorld(model_name)
             model.set_classes(labels)
-        else:
-            logger.warning("The model does not support setting classes directly.")
-    elif backend == PredictorBackend.robotoff:
-        if server_url is None:
-            raise typer.BadParameter("--server-url is required for Robotoff backend")
+
+    elif backend == PredictorBackend.ultralytics_sam3:
+        from ultralytics.models.sam import SAM3SemanticPredictor
 
         if threshold is None:
-            threshold = 0.1
-            server_url = server_url.rstrip("/")
+            threshold = 0.25
+
+        # SAM3 cannot be downloaded directly using to to a gated access. Use a
+        # proxy repo.
+        model_path = hf_hub_download(
+            "1038lab/sam3",
+            filename="sam3.pt",
+            revision="f055b060a4de0a040891ba2ebac9c5cb3c1c0132",
+        )
+        overrides = dict(
+            task="segment",
+            mode="predict",
+            model=model_path,
+            save=False,
+        )
+
+        if imgsz is not None:
+            overrides["imgsz"] = imgsz
+        model = SAM3SemanticPredictor(overrides=overrides)
     else:
         raise typer.BadParameter(f"Unsupported backend: {backend}")
 
     for task in tqdm.tqdm(
         ls.tasks.list(project=project_id, view=view_id), desc="tasks"
     ):
-        if task.total_predictions == 0:
+        if not (skip_existing and task.total_predictions > 0):
             image_url = task.data["image_url"]
-            image = typing.cast(
-                Image.Image,
-                get_image_from_url(image_url, error_raise=error_raise),
-            )
-            if backend == PredictorBackend.ultralytics:
-                results = model.predict(
-                    image,
-                    conf=threshold,
-                    max_det=max_det,
-                )[0]
+            image = get_image_from_url(image_url, error_raise=error_raise)
+            if image is None:
+                continue
+            min_score = None
+            if backend in (
+                PredictorBackend.ultralytics,
+                PredictorBackend.ultralytics_yolo_world,
+            ):
+                predict_kwargs = {
+                    "conf": threshold,
+                    "max_det": max_det,
+                }
+                if imgsz is not None:
+                    predict_kwargs["imgsz"] = imgsz
+                results = model.predict(image, **predict_kwargs)[0]
                 labels = typing.cast(list[str], labels)
                 label_studio_result = format_annotation_results_from_ultralytics(
                     results, labels, label_mapping_dict
                 )
-            elif backend == PredictorBackend.robotoff:
-                r = http_session.get(
-                    f"{server_url}/api/v1/images/predict",
-                    params={
-                        "models": model_name,
-                        "output_image": 0,
-                        "image_url": image_url,
-                    },
-                )
-                r.raise_for_status()
-                response = r.json()
-                label_studio_result = format_annotation_results_from_robotoff(
-                    response["predictions"][model_name],
-                    image.width,
-                    image.height,
-                    label_mapping_dict,
+            elif backend == PredictorBackend.ultralytics_sam3:
+                model.set_image(image)
+                results = model(text=labels)[0]
+                label_studio_result = format_annotation_results_from_ultralytics(
+                    results, labels, label_mapping_dict
                 )
+                min_score = min(results.boxes.conf.tolist(), default=None)
             if dry_run:
                 logger.info("image_url: %s", image_url)
                 logger.info("result: %s", label_studio_result)
@@ -372,7 +482,9 @@ def add_prediction(
                     task=task.id,
                     result=label_studio_result,
                     model_version=model_version,
+                    score=min_score,
                 )
+                logger.info("Prediction added for task: %s", task.id)
 
 
 @app.command()
@@ -449,28 +561,61 @@ def create_config_file(
 @app.command()
 def check_dataset(
     project_id: Annotated[int, typer.Option(help="Label Studio Project ID")],
+    view_id: Annotated[int, typer.Option(help="Label Studio View ID, if any.")] = None,
     api_key: Annotated[
-        Optional[str], typer.Option(envvar="LABEL_STUDIO_API_KEY")
-    ] = None,
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
+    delete_missing_images: Annotated[
+        bool,
+        typer.Option(help="Delete tasks with missing images from the dataset"),
+    ] = False,
+    delete_duplicate_images: Annotated[
+        bool, typer.Option(help="Delete duplicate images from the dataset")
+    ] = False,
 ):
-    """Check a dataset for duplicate images on Label Studio."""
+    """Perform sanity checks of a Label Studio dataset.
+
+    This function checks for:
+    - Tasks with missing images (404)
+    - Duplicate images based on perceptual hash (pHash)
+    - Tasks with multiple annotations
+
+    This function doesn't perform any modifications to the dataset, except
+    optionally deleting tasks with missing images if --delete-missing-images
+    is provided and tasks with duplicate images if --delete-duplicate-images
+    is provided.
+    """
     from label_studio_sdk.client import LabelStudio
 
     from ..check import check_ls_dataset
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
-    check_ls_dataset(ls, project_id)
+    check_ls_dataset(
+        ls=ls,
+        project_id=project_id,
+        view_id=view_id,
+        delete_missing_images=delete_missing_images,
+        delete_duplicate_images=delete_duplicate_images,
+    )
 
 
 @app.command()
 def list_users(
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
 ):
     """List all users in Label Studio."""
     from label_studio_sdk.client import LabelStudio
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
 
     for user in ls.users.list():
@@ -480,11 +625,57 @@ def list_users(
 @app.command()
 def delete_user(
     user_id: int,
-    api_key: Annotated[str, typer.Option(envvar="LABEL_STUDIO_API_KEY")],
-    label_studio_url: str = LABEL_STUDIO_DEFAULT_URL,
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
 ):
     """Delete a user from Label Studio."""
     from label_studio_sdk.client import LabelStudio
 
+    check_label_studio_api_key(api_key)
     ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
     ls.users.delete(user_id)
+
+
+@app.command()
+def dump_dataset(
+    project_id: Annotated[int, typer.Option(help="Label Studio Project ID")],
+    output_file: Annotated[
+        Path, typer.Option(help="Path of the output file", writable=True)
+    ],
+    api_key: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_API_KEY)
+    ] = config.label_studio_api_key,
+    view_id: Annotated[
+        int | None,
+        typer.Option(
+            help="ID of the Label Studio view, if any. This option is useful "
+            "to filter the tasks to dump."
+        ),
+    ] = None,
+    label_studio_url: Annotated[
+        str, typer.Option(help=typer_description.LABEL_STUDIO_URL)
+    ] = config.label_studio_url,
+):
+    """Dump all the tasks of a dataset in a JSONL file.
+
+    All fields of the tasks are exported. A subset of the tasks can be
+    selected by filtering tasks based on a view (=tab) using the `--view-id`
+    option.
+    """
+    import orjson
+    import tqdm
+    from label_studio_sdk.client import LabelStudio
+
+    check_label_studio_api_key(api_key)
+    ls = LabelStudio(base_url=label_studio_url, api_key=api_key)
+
+    with output_file.open("wb") as f:
+        for task in tqdm.tqdm(
+            ls.tasks.list(project=project_id, view=view_id), desc="tasks"
+        ):
+            content = orjson.dumps(task.dict())
+            f.write(content + b"\n")

labelr/apps/typer_description.py

@@ -0,0 +1,2 @@
+LABEL_STUDIO_API_KEY = """API Key to authenticate to the Label Studio server. Can also be set with the LABELR_LABEL_STUDIO_API_KEY environment variable."""
+LABEL_STUDIO_URL = """URL of the Label Studio server. Can also be set with the LABELR_LABEL_STUDIO_URL environment variable."""