Merge pull request #180 from fastlabel/feature/dataset-objects-api

feat: dataset objects api

Author: ueno-k-work

README.md

@@ -2477,11 +2477,41 @@ Create object in the dataset.
 The types of objects that can be created are "image", "video", and "audio".
 There are type-specific methods. but they can be used in the same way.
 
+Created object are automatically assigned to the "latest" dataset version.
+
 ```python
 dataset_object = client.create_dataset_object(
-    dataset_version_id="YOUR_DATASET_VERSION_ID",
+    dataset="YOUR_DATASET_NAME",
     name="brushwood_dog.jpg",
     file_path="./brushwood_dog.jpg",
+    tags=["dog"], # max 5 tags per dataset object.
+    annotations=[
+        {
+            "keypoints": [
+                {
+                    "value": [
+                        102.59,
+                        23.04,
+                        1
+                    ],
+                    "key": "head"
+                }
+            ],
+            "attributes": [
+                {
+                    "value": "Scottish field",
+                    "key": "kind"
+                }
+            ],
+            "confidenceScore": 0,
+            "rotation": 0,
+            "points": [
+                0
+            ],
+            "value": "dog",
+            "type": "bbox" # type can be 'bbox', 'segmentation'.
+        }
+    ]
 )
 ```
 
@@ -2496,7 +2526,46 @@ See API docs for details.
     'size': 6717,
     'height': 225,
     'width': 225,
-    'groupId': None,
+    'tags': [
+        'dog'
+    ],
+    "annotations": [
+        {
+            "id": "YOUR_DATASET_OBJECT_ANNOTATION_ID",
+            "type": "bbox",
+            "title": "dog",
+            "value": "dog",
+            "points": [
+                0
+            ],
+            "attributes": [
+                {
+                    "value": "Scottish field",
+                    "key": "kind",
+                    "name": "Kind",
+                    "type": "text"
+                }
+            ],
+            "keypoints": [
+                {
+                    "edges": [
+                        "right_shoulder",
+                        "left_shoulder"
+                    ],
+                    "value": [
+                        102.59,
+                        23.04,
+                        1
+                    ],
+                    "key": "head",
+                    "name": "頭"
+                }
+            ],
+            "rotation": 0,
+            "color": "#FF0000",
+            "confidenceScore": -1
+        }
+    ],
     'createdAt': '2022-10-30T08:32:20.748Z',
     'updatedAt': '2022-10-30T08:32:20.748Z'
 }
@@ -2519,20 +2588,28 @@ Success response is the same as when created.
 Get all dataset object in the dataset. (Up to 1000 tasks)
 
 ```python
-dataset_objects = client.get_dataset_objects(dataset_version_id="YOUR_DATASET_VERSION_ID")
+dataset_objects = client.get_dataset_objects(dataset="YOUR_DATASET_NAME")
 ```
 
 The success response is the same as when created, but it is an array.
 
-You can filter by keywords.
+You can filter by version and tags.
 
 ```python
 dataset_objects = client.get_dataset_objects(
-    dataset_version_id="YOUR_DATASET_VERSION_ID", keyword="dog"
+    dataset="YOUR_DATASET_NAME",
+    version="latest", # default is "latest"
+    tags=["cat"],
 )
 ```
 
-If you wish to retrieve more than 1000 data sets, please refer to the Task [sample code](#get-tasks).
+### Delete Dataset Object
+
+Delete a single dataset object.
+
+```python
+client.delete_dataset_object(dataset_object_id="YOUR_DATASET_OBJECT_ID")
+```
 
 ## Converter
 

examples/create_dataset_object.py

@@ -5,7 +5,7 @@
 client = fastlabel.Client()
 
 dataset_object = client.create_dataset_object(
-    dataset_version_id="YOUR_DATASET_VERSION_ID",
+    dataset="YOUR_DATASET_NAME",
     name="NAME",
     file_path="FILE_PATH",
 )

examples/delete_dataset_object.py

@@ -0,0 +1,5 @@
+import fastlabel
+
+client = fastlabel.Client()
+
+client.delete_dataset_object(dataset_object_id="YOUR_DATASET_OBJECT_ID")

examples/get_dataset_objects.py

@@ -4,7 +4,5 @@
 
 client = fastlabel.Client()
 
-dataset_objects = client.get_dataset_objects(
-    dataset_version_id="YOUR_DATASET_VERSION_ID"
-)
+dataset_objects = client.get_dataset_objects(dataset="YOUR_DATASET_NAME")
 pprint(dataset_objects)

fastlabel/__init__.py

@@ -3910,58 +3910,65 @@ def find_dataset_object(self, dataset_object_id: str) -> dict:
 
     def get_dataset_objects(
         self,
-        dataset_version_id: str,
-        keyword: str = None,
-        offset: int = None,
-        limit: int = 100,
+        dataset: str,
+        version: str = None,
+        tags: List[str] = [],
     ) -> list:
         """
         Returns a list of dataset objects.
 
-        Returns up to 1000 at a time, to get more, set offset as the starting position
-        to fetch.
-
-        dataset_version_id is dataset object in dataset version (Required).
-        keyword are search terms in the dataset object name (Optional).
-        offset is the starting position number to fetch (Optional).
-        limit is the max number to fetch (Optional).
+        dataset is dataset name (Required).
+        version is dataset version (Optional).
+        tags is a list of tag (Optional).
         """
-        if limit > 1000:
-            raise FastLabelInvalidException(
-                "Limit must be less than or equal to 1000.", 422
-            )
         endpoint = "dataset-objects"
-        params = {"datasetVersionId": dataset_version_id}
-        if keyword:
-            params["keyword"] = keyword
-        if offset:
-            params["offset"] = offset
-        if limit:
-            params["limit"] = limit
+        params = {"dataset": dataset}
+        if version:
+            params["version"] = version
+        if tags:
+            params["tags"] = tags
         return self.api.get_request(endpoint, params=params)
 
     def create_dataset_object(
-        self, dataset_version_id: str, name: str, file_path: str
+        self,
+        dataset: str,
+        name: str,
+        file_path: str,
+        tags: List[str] = [],
+        annotations: List[dict] = [],
     ) -> dict:
         """
         Create a dataset object.
 
-        dataset_version_id is dataset object in dataset version (Required).
+        dataset is dataset name (Required).
         name is a unique identifier of dataset object in your dataset (Required).
         file_path is a path to data. (Required).
+        tags is a list of tag (Optional).
+        annotations is a list of annotation (Optional).
         """
         endpoint = "dataset-objects"
         if not utils.is_object_supported_size(file_path):
             raise FastLabelInvalidException(
                 "Supported object size is under 250 MB.", 422
             )
         payload = {
-            "datasetVersionId": dataset_version_id,
+            "dataset": dataset,
             "name": name,
-            "file": utils.base64_encode(file_path),
+            "filePath": utils.base64_encode(file_path),
         }
+        if tags:
+            payload["tags"] = tags
+        if annotations:
+            payload["annotations"] = annotations
         return self.api.post_request(endpoint, payload=payload)
 
+    def delete_dataset_object(self, dataset_object_id: str) -> None:
+        """
+        Delete a dataset object.
+        """
+        endpoint = "dataset-objects/" + dataset_object_id
+        self.api.delete_request(endpoint)
+
     def update_aws_s3_storage(
         self, project: str, bucket_name: str, bucket_region: str, prefix: str = None
     ) -> str:

fastlabel/api.py

@@ -1,12 +1,12 @@
 import os
-import requests
 from typing import Union
 
+import requests
+
 from .exceptions import FastLabelException, FastLabelInvalidException
 
 
 class Api:
-
     base_url = "https://api.fastlabel.ai/v1/"
     access_token = None
 
@@ -124,6 +124,5 @@ def upload_zipfile(
         url: str,
         file_path: str,
     ):
-        files = {'file': open(file_path, 'rb')}
+        files = {"file": open(file_path, "rb")}
         return requests.put(url, files=files)
-