feat: support FastLabel SDK (#169)

* add fastlabel sdk to poetry

Signed-off-by: Shunsuke Miura <[email protected]>

* feat: add fastlabel label downloader

Signed-off-by: Shunsuke Miura <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove unnecessary blank

Signed-off-by: Shunsuke Miura <[email protected]>

* update to output all and each task

Signed-off-by: Shunsuke Miura <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix bug

Signed-off-by: Shunsuke Miura <[email protected]>

* update to use config file

Signed-off-by: Shunsuke Miura <[email protected]>

* add access token to config file

Signed-off-by: Shunsuke Miura <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* modifle gitignore, remove label file

Signed-off-by: Shunsuke Miura <[email protected]>

* fix output_dir name

Signed-off-by: Shunsuke Miura <[email protected]>

* fix anonymizxation_dir

Signed-off-by: Shunsuke Miura <[email protected]>

---------

Signed-off-by: Shunsuke Miura <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

.gitignore

@@ -4,6 +4,7 @@ tests/scale_to_t4/data_test_scale_to_t4_converter/output_base
 /data*/
 /notebooks/data/
 /tests/data
+/label_data*
 
 # ros
 /src

config/download_fastlabel_annotations.yaml

@@ -0,0 +1,37 @@
+task: donwload_fastlabel
+conversion:
+  output: label_data
+  access_token:
+  target_project_slugs:
+    - 202405-lidar-3d-bbox-trial
+    - 202405-lidar-3d-bbox-trial-2
+    - 202405-lidar-3d-bbox-trial-3
+    - 202405-lidar-3d-bbox-trial-4
+    - 202405-panoptic-segmentation-trial
+    - 202405-panoptic-segmentation-trial-2
+    - 202405-panoptic-segmentation-trial-3
+    - 202405-panoptic-segmentation-trial-4
+    - 202405-license-plate-person-head-bbox-trial
+    - 202405-traffic-light-recognition-trial
+    - 3d-lidar-bbox-202409-01
+    - 3d-lidar-bbox-202409-02
+    - 3d-lidar-bbox-202410-lo-01
+    - 3d-lidar-bbox-202410-lo-02
+    - 3d-lidar-bbox-202410-ar-01
+    - 3d-lidar-bbox-202410-ar-02
+    - 3d-lidar-bbox-202411-lo-01
+    - 3d-lidar-bbox-202411-lo-02
+    - 3d-lidar-bbox-202411-ar-01
+    - panoptic-segmentation-202409-01
+    - panoptic-segmentation-202409-02
+    - panoptic-segmentation-202410-lo-01
+    - panoptic-segmentation-202410-lo-02
+    - panoptic-segmentation-202410-ar-01
+    - panoptic-segmentation-202410-ar-02
+    - panoptic-segmentation-202411-lo-01
+    - panoptic-segmentation-202411-lo-02
+    - panoptic-segmentation-202411-ar-01
+    - traffic-light-recognition-202410
+    - traffic-light-recognition-202409-01
+    - license-plate-person-head-202410
+    - license-plate-person-head-202409-01

perception_dataset/fastlabel/download_annotations.py

@@ -0,0 +1,177 @@
+import argparse
+import json
+import os
+import os.path as osp
+import time
+from typing import Dict, List
+
+import fastlabel
+import yaml
+
+
+def load_yaml_config(yaml_path: str) -> Dict:
+    with open(yaml_path, "r") as f:
+        return yaml.safe_load(f)
+
+
+def get_large_size_tasks(project: Dict, limit: int = None) -> List[Dict]:
+    # Iterate pages until new tasks are empty.
+    large_size_tasks = []
+    offset = 0
+    while True:
+        time.sleep(1)
+
+        if limit is not None:
+            donwload_limit = limit
+        else:
+            donwload_limit = 100
+        print(f"Downloading label: {project['slug']}, offset: {offset:05d}. Please wait...")
+        if "image" in project["type"]:
+            tasks = client.get_image_tasks(
+                project=project["slug"], offset=offset, limit=donwload_limit
+            )
+        elif "pcd" in project["type"]:
+            tasks = client.get_pcd_tasks(
+                project=project["slug"], offset=offset, limit=donwload_limit
+            )
+        else:
+            raise ValueError(f"Unknown project type: {project['type']}")
+        large_size_tasks.extend(tasks)
+
+        if limit is not None and len(large_size_tasks) >= limit:
+            break
+        elif len(tasks) > 0:
+            offset = len(large_size_tasks)  # Set the offset
+        else:
+            break
+    return large_size_tasks
+
+
+def split_project_pcd_image(projects: List[Dict]) -> Dict[str, List[Dict]]:
+    cuboid_projects = []
+    seg_projects = []
+    bbox_projects = []
+
+    for project in projects:
+        if "pcd_cuboid" in project["type"]:
+            cuboid_projects.append(project)
+        elif "segmentation" in project["type"] or project["type"] == "image_all":
+            seg_projects.append(project)
+        elif "bbox" in project["type"]:
+            bbox_projects.append(project)
+    return {"cuboid": cuboid_projects, "segmentation": seg_projects, "bbox": bbox_projects}
+
+
+def is_task_completed(task: Dict) -> bool:
+    return task["status"] in ["approved", "completed"]
+
+
+def rename_image_task_name(name: str) -> str:
+    if "_CAM_" not in name:
+        return name
+    base_name, camera_info = name.split("CAM_")
+    dataset_name, index = base_name.split("/")
+    frame_index = index[:5]
+    camera_name, ext = os.path.splitext(camera_info)
+    new_file_name = f"{dataset_name}/CAM_{camera_name}/{frame_index}{ext}"
+    return new_file_name
+
+
+def download_completed_annotations(
+    project: Dict, output_dir: str, target_project_slugs: List[str], save_each: bool = False
+) -> None:
+    os.makedirs(output_dir, exist_ok=True)
+    completed_tasks = []
+    if project["slug"] in target_project_slugs:
+        tasks = get_large_size_tasks(project=project)
+        for task in tasks:
+            if is_task_completed(task):
+                task["name"] = rename_image_task_name(task["name"])
+                completed_tasks.append(task)
+                each_task_file_name = task["name"].replace(" ", "_").replace("/", "_") + ".json"
+                if save_each:
+                    with open(osp.join(output_dir, each_task_file_name), "w") as f:
+                        json.dump([task], f, indent=4)
+            else:
+                print(f"Task {task['name']} is not completed. Removing.")
+    return completed_tasks
+
+
+def get_labels(output_dir: str, target_project_slugs: List[str]) -> None:
+    projects = client.get_projects()
+    project_dict = split_project_pcd_image(projects)
+    os.makedirs(output_dir, exist_ok=True)
+
+    cuboid_tasks = []
+    pcd_out_dir = osp.join(output_dir, "pcd_annotation")
+    for project in project_dict["cuboid"]:
+        if project["slug"] in target_project_slugs:
+            cuboid_tasks.extend(
+                download_completed_annotations(
+                    project, pcd_out_dir, target_project_slugs, save_each=True
+                )
+            )
+    with open(osp.join(output_dir, "all_label_lidar_cuboid.json"), "w") as f:
+        json.dump(cuboid_tasks, f, indent=4)
+
+    segmentation_tasks = []
+    pcd_out_dir = osp.join(output_dir, "segmentation_annotation")
+    for project in project_dict["segmentation"]:
+        if project["slug"] in target_project_slugs:
+            segmentation_tasks.extend(
+                download_completed_annotations(
+                    project, pcd_out_dir, target_project_slugs, save_each=True
+                )
+            )
+    with open(osp.join(output_dir, "all_label_segmentation.json"), "w") as f:
+        json.dump(segmentation_tasks, f, indent=4)
+
+    tlr_tasks = []
+    pcd_out_dir = osp.join(output_dir, "tlr_annotation")
+    for project in project_dict["bbox"]:
+        if project["slug"] in target_project_slugs and "traffic-light" in project["slug"]:
+            tlr_tasks.extend(
+                download_completed_annotations(
+                    project, pcd_out_dir, target_project_slugs, save_each=True
+                )
+            )
+    with open(osp.join(output_dir, "all_label_traffic_light.json"), "w") as f:
+        json.dump(tlr_tasks, f, indent=4)
+
+    anonymization_tasks = []
+    pcd_out_dir = osp.join(output_dir, "anonymization")
+    for project in project_dict["bbox"]:
+        if project["slug"] in target_project_slugs and "license-plate" in project["slug"]:
+            anonymization_tasks.extend(
+                download_completed_annotations(
+                    project, pcd_out_dir, target_project_slugs, save_each=True
+                )
+            )
+    with open(osp.join(output_dir, "all_label_anonymization.json"), "w") as f:
+        json.dump(anonymization_tasks, f, indent=4)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--config",
+        type=str,
+        required=True,
+        help="Path to the YAML configuration file.",
+    )
+    args = parser.parse_args()
+
+    config = load_yaml_config(args.config)
+
+    output_dir = config.get("conversion", {}).get("output", "label_data")
+    target_project_slugs = config.get("conversion", {}).get("target_project_slugs", [])
+    access_token = config.get("conversion", {}).get("access_token", "")
+    if access_token is not None and access_token != "":
+        os.environ["FASTLABEL_ACCESS_TOKEN"] = access_token
+    client = fastlabel.Client()
+
+    if not target_project_slugs:
+        print("No target project slugs found in the YAML file. Exiting.")
+        exit(1)
+
+    get_labels(output_dir, target_project_slugs)