feat: add scene and frame describe

Author: ankit-v2-3

videodb/_constants.py

@@ -27,6 +27,10 @@ class SceneExtractionType:
     time_based = "time"
 
 
+class SceneModels:
+    gpt4_o = "gpt4-o"
+
+
 class Workflows:
     add_subtitles = "add_subtitles"
 
@@ -58,6 +62,8 @@ class ApiPath:
     invoices = "invoices"
     scenes = "scenes"
     scene = "scene"
+    frame = "frame"
+    describe = "describe"
 
 
 class Status:

videodb/image.py

@@ -1,5 +1,6 @@
 from videodb._constants import (
     ApiPath,
+    SceneModels,
 )
 
 
@@ -61,3 +62,11 @@ def to_json(self):
             "frame_time": self.frame_time,
             "description": self.description,
         }
+
+    def describe(self, prompt: str = None, model_name=SceneModels.gpt4_o):
+        description_data = self._connection.post(
+            path=f"{ApiPath.video}/{self.video_id}/{ApiPath.frame}/{self.id}/{ApiPath.describe}",
+            data={"prompt": prompt, "model_name": model_name},
+        )
+        self.description = description_data.get("description", None)
+        return self.description

videodb/scene.py

@@ -1,6 +1,6 @@
 from typing import List
 
-from videodb._constants import ApiPath
+from videodb._constants import ApiPath, SceneModels
 
 from videodb.image import Frame
 
@@ -14,13 +14,15 @@ def __init__(
         description: str,
         id: str = None,
         frames: List[Frame] = [],
+        connection=None,
     ):
         self.id = id
         self.video_id = video_id
         self.start = start
         self.end = end
         self.frames: List[Frame] = frames
         self.description = description
+        self._connection = connection
 
     def __repr__(self) -> str:
         return (
@@ -43,6 +45,16 @@ def to_json(self):
             "description": self.description,
         }
 
+    def describe(self, prompt: str = None, model_name=SceneModels.gpt4_o) -> None:
+        if self._connection is None:
+            raise ValueError("Connection is required to describe a scene")
+        description_data = self._connection.post(
+            path=f"{ApiPath.video}/{self.video_id}/{ApiPath.scene}/{self.id}/{ApiPath.describe}",
+            data={"prompt": prompt, "model_name": model_name},
+        )
+        self.description = description_data.get("description", None)
+        return self.description
+
 
 class SceneCollection:
     def __init__(

videodb/video.py

@@ -197,6 +197,7 @@ def _format_scene_collection(self, scene_collection_data: dict) -> SceneCollecti
                 description=scene.get("description"),
                 id=scene.get("scene_id"),
                 frames=frames,
+                connection=self._connection,
             )
             scenes.append(scene)