river_abnormal_detect

This project is designed to detect abnormalities in river environments by analyzing video streams. It uses advanced techniques such as adaptive background subtraction, flow analysis, and object detection to classify detected objects and events.

Prerequisites

1. Install conda.
2. Use the requirements.txt file to set up the environment.

Environment Setup

conda create --name river_abnormal_detect --file requirements.txt
conda activate river_abnormal_detect

Modes of Operation

1. Offline Testing

• Run the video_detect.py script:

  python video_detect.py

• Modify the video_path variable in video_detect.py to the path of the video file you want to test.

2. Online Testing

• Run the streaming_detect.py script:

  python streaming_detect.py

• Modify the youtube_url variable in streaming_detect.py to include the URL of the video stream.

Real-Time Upload of Predictions and Video Captures

There are two options for uploading predictions and captured videos in real time:

Option 1: Telegram Bot

1. Update the tg_token, tg_chat_id field in utils/config_detect.py with your Telegram bot's API key.
2. When running detection, you must also execute the utils/telegram_bot.py script:

   python utils/telegram_bot.py

Option 2: LINE Bot

1. Update the LineBotApi("") & user_id field in utils/config_detect.py with your LINE bot's API key.