update examples

camel/toolkits/video_toolkit.py

@@ -17,7 +17,7 @@
 import re
 import tempfile
 from pathlib import Path
-from typing import List, Optional, Union
+from typing import List, Optional
 
 from PIL import Image
 
@@ -41,12 +41,12 @@ def _standardize_url(url: str) -> str:
     return url
 
 
-def _capture_screenshot(video_file: str, timestamp: int) -> Image.Image:
+def _capture_screenshot(video_file: str, timestamp: float) -> Image.Image:
     r"""Capture a screenshot from a video file at a specific timestamp.
 
     Args:
         video_file (str): The path to the video file.
-        timestamp (int): The time in seconds from which to capture the
+        timestamp (float): The time in seconds from which to capture the
           screenshot.
 
     Returns:
@@ -93,8 +93,6 @@ def __init__(
             download_directory or tempfile.mkdtemp()
         ).resolve()
 
-        logger.info(f"self._download_directory: {self._download_directory}")
-
         try:
             self._download_directory.mkdir(parents=True, exist_ok=True)
         except FileExistsError:
@@ -133,7 +131,7 @@ def _download_video(self, url: str) -> str:
         video_template = self._download_directory / "%(title)s.%(ext)s"
         ydl_opts = {
             'format': 'bestvideo+bestaudio/best',
-            'outtmpl': video_template,
+            'outtmpl': str(video_template),
             'force_generic_extractor': True,
             'cookiefile': self._cookies_path,
         }
@@ -168,16 +166,14 @@ def get_video_bytes(
         return video_bytes
 
     def get_video_screenshots(
-        self, video_url: str, timestamps: Union[List[int], int]
+        self, video_url: str, amount: int
     ) -> List[Image.Image]:
         r"""Capture screenshots from the video at specified timestamps or by
         dividing the video into equal parts if an integer is provided.
 
         Args:
-            video_url (str): The URL of the video to download. (required)
-            timestamps (Union[List[int], int]): A list of timestamps (in
-                seconds) from which to capture the screenshots, or an integer
-                specifying the number of evenly spaced screenshots to capture.
+            video_url (str): The URL of the video to take screenshots.
+            amount (int): the amount of evenly split screenshots to capture.
 
         Returns:
             List[Image.Image]: A list of screenshots as Image.Image.
@@ -194,17 +190,10 @@ def get_video_screenshots(
         except ffmpeg.Error as e:
             raise RuntimeError(f"Failed to determine video length: {e.stderr}")
 
-        if isinstance(timestamps, int):
-            if timestamps <= 0:
-                raise ValueError(
-                    "Number of screenshots must be greater than zero."
-                )
-            interval = video_length // (timestamps + 1)
-            tss = [round((i + 1) * interval) for i in range(timestamps)]
-        else:
-            tss = [ts for ts in timestamps if 0 <= ts <= video_length]
+        interval = video_length / (amount + 1)
+        timestamps = [i * interval for i in range(1, amount + 1)]
 
-        images = [_capture_screenshot(video_file, ts) for ts in tss]
+        images = [_capture_screenshot(video_file, ts) for ts in timestamps]
 
         return images
 

examples/vision/duckduckgo_video_object_recognition.py

@@ -11,69 +11,59 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # =========== Copyright 2023 @ CAMEL-AI.org. All Rights Reserved. ===========
+from typing import List
+
+from PIL import Image
+
 from camel.agents import ChatAgent
-from camel.configs import ChatGPTConfig
 from camel.messages import BaseMessage
-from camel.models import ModelFactory
 from camel.prompts import PromptTemplateGenerator
 from camel.toolkits import SearchToolkit, VideoDownloaderToolkit
-from camel.types import ModelPlatformType, ModelType, RoleType, TaskType
+from camel.types import RoleType, TaskType
 
 
-def detect_image_obj(image_list) -> None:
+def detect_image_obj(image_list: List[Image.Image]) -> None:
     sys_msg = PromptTemplateGenerator().get_prompt_from_key(
         TaskType.OBJECT_RECOGNITION, RoleType.ASSISTANT
     )
     print("=" * 20 + " SYS MSG " + "=" * 20)
     print(sys_msg)
     print("=" * 49)
-
-    assistant_sys_msg = BaseMessage.make_assistant_message(
-        role_name="Assistant",
-        content=sys_msg,
-    )
-    model = ModelFactory.create(
-        model_platform=ModelPlatformType.OPENAI,
-        model_type=ModelType.GPT_4O_MINI,
-        model_config_dict=ChatGPTConfig().as_dict(),
-    )
-    agent = ChatAgent(
-        assistant_sys_msg,
-        model=model,
-    )
+    agent = ChatAgent(sys_msg)
 
     user_msg = BaseMessage.make_user_message(
         role_name="User",
         content="Please start the object detection for the following images!",
         image_list=image_list,
         image_detail="high",
     )
+
     assistant_response = agent.step(user_msg)
     print("=" * 20 + " RESULT " + "=" * 20)
     print(assistant_response.msgs[0].content)
     print("=" * 48)
 
 
-# Create an instance of the SearchToolkit
-search_toolkit = SearchToolkit()
+def main():
+    # Create an instance of the SearchToolkit
+    search_toolkit = SearchToolkit()
 
-# Example query for DuckDuckGo video search
-query = "What big things are happening in 2024?"
+    # Example query for DuckDuckGo video search
+    query = "The future of AI in education"
 
-# Perform a DuckDuckGo search with the query, setting source to 'videos'
-results = search_toolkit.search_duckduckgo(
-    query=query, source="videos", max_results=5
-)
+    # Perform a DuckDuckGo search with the query, setting source to 'videos'
+    results = search_toolkit.search_duckduckgo(
+        query=query, source="videos", max_results=5
+    )
 
-# Try to download videos from the search results
-for result in results:
-    video_url = result['embed_url']
-    if not video_url:
-        print(f"No valid video URL provided for result: {result}")
-        continue
+    # Try to download videos from the search results
+    for result in results:
+        video_url = result['embed_url']
+        if not video_url:
+            print(f"No valid video URL provided for result: {result}")
+            continue
 
-    print(f"Trying to download video from: {video_url}")
-    try:
+        print(f"Trying to download video from: {video_url}")
         downloader = VideoDownloaderToolkit()
         image_list = downloader.get_video_screenshots(video_url, 3)
         if image_list and len(image_list) > 0:
@@ -86,10 +76,13 @@ def detect_image_obj(image_list) -> None:
             break
         else:
             print(f"Failed to capture screenshots from video: {video_url}")
-    except Exception as e:
-        print(f"Failed to download video from {video_url}: {e!s}")
 
-print("Exited the video download loop.")
+    print("Exited the video download loop.")
+
+
+if __name__ == "__main__":
+    main()
+
 """
 ===============================================================================
 Successfully downloaded video and captured screenshots 

examples/vision/web_video_description_extractor.py

@@ -12,36 +12,24 @@
 # limitations under the License.
 # =========== Copyright 2023 @ CAMEL-AI.org. All Rights Reserved. ===========
 from camel.agents import ChatAgent
-from camel.configs import ChatGPTConfig
 from camel.messages import BaseMessage
-from camel.models import ModelFactory
 from camel.prompts import PromptTemplateGenerator
 from camel.toolkits import VideoDownloaderToolkit
-from camel.types import ModelPlatformType, ModelType, RoleType, TaskType
+from camel.types import RoleType, TaskType
 
 video_url = (
-    'https://sample-videos.com/video321/mp4/720/big_buck_bunny_720p_1mb.mp4'
+    "https://sample-videos.com/video321/mp4/720/big_buck_bunny_720p_1mb.mp4"
 )
 downloader = VideoDownloaderToolkit()
 
 # Get the video bytes
 video_bytes = downloader.get_video_bytes(video_url)
 
-sys_msg_prompt = PromptTemplateGenerator().get_prompt_from_key(
+sys_msg = PromptTemplateGenerator().get_prompt_from_key(
     TaskType.VIDEO_DESCRIPTION, RoleType.ASSISTANT
 )
-sys_msg = BaseMessage.make_assistant_message(
-    role_name="Assistant",
-    content=sys_msg_prompt,
-)
-
-model = ModelFactory.create(
-    model_platform=ModelPlatformType.OPENAI,
-    model_type=ModelType.GPT_4O,
-    model_config_dict=ChatGPTConfig().as_dict(),
-)
 
-camel_agent = ChatAgent(sys_msg, model=model)
+camel_agent = ChatAgent(sys_msg)
 
 # Create user message with video bytes
 user_msg = BaseMessage.make_user_message(

examples/vision/web_video_object_recognition.py

@@ -14,14 +14,10 @@
 import argparse
 
 from camel.agents import ChatAgent
-from camel.configs.openai_config import ChatGPTConfig
 from camel.generators import PromptTemplateGenerator
 from camel.messages import BaseMessage
-from camel.models import ModelFactory
 from camel.toolkits.video_toolkit import VideoDownloaderToolkit
 from camel.types import (
-    ModelPlatformType,
-    ModelType,
     RoleType,
     TaskType,
 )
@@ -49,19 +45,7 @@ def detect_image_obj(image_list) -> None:
     print(sys_msg)
     print("=" * 49)
 
-    assistant_sys_msg = BaseMessage.make_assistant_message(
-        role_name="Assistant",
-        content=sys_msg,
-    )
-    model = ModelFactory.create(
-        model_platform=ModelPlatformType.OPENAI,
-        model_type=ModelType.GPT_4O_MINI,
-        model_config_dict=ChatGPTConfig().as_dict(),
-    )
-    agent = ChatAgent(
-        assistant_sys_msg,
-        model=model,
-    )
+    agent = ChatAgent(sys_msg)
 
     user_msg = BaseMessage.make_user_message(
         role_name="User",