Sharp Object Detection Alert

Project Overview

The Sharp Object Detection Alert is a cutting-edge tool designed to detect sharp objects in videos. It processes videos provided via Google Drive links, analyzes frames using the YOLOv8 model (yolov8n.pt), sends the frames to Azure Computer Vision for object detection, and then utilizes an OpenAI-powered assistant to evaluate the results. Based on the analysis, the system sends alert, result, or error notifications via Gmail SMTP.

Features

• Video frame extraction and analysis using Python-OpenCV and YOLOv8 model.
• Integration with Azure Computer Vision for object detection.
• OpenAI Assistant for interpreting the results from Computer Vision.
• Email notifications sent using Gmail SMTP:
  • Alert email: When a sharp object is detected.
  • Result notification: When no sharp object is detected.
  • Error notification: When an error occurs during processing.

Services Integrated

1. YOLOv8 Model (yolov8n.pt)

• The YOLOv8 model is used for detecting objects in each video frame. It helps identify potential sharp objects based on the features trained into the model.

2. Azure Computer Vision

• Once frames are extracted, they are sent to Azure Computer Vision for object content and object detection. It analyzes the image and provides a detailed result (objects detected, confidence score, etc.).

3. OpenAI Assistant

• The OpenAI assistant evaluates the JSON response from Azure Computer Vision, interpreting whether a sharp object is present, what object it is, and providing additional details about the object detected.

4. Gmail SMTP Service

• The system sends emails via Gmail SMTP, using the results from the analysis. There are three types of emails:
  • Alert Email: If a sharp object is detected in the video.
  • Result Notification: If no sharp object is detected.
  • Error Notification: If any error occurs during the video processing.

Getting Started

Prerequisites

Make sure you have the following installed on your machine:

• Python 3.12: Ensure you have Python installed by running:

  python --version

• Google Drive link for the video to be analyzed following the pattern https://drive.google.com/uc?id=ID_GOOGLE_DRIVE_VIDEO.
• Email account for sending emails via SMTP.

Installation Steps

Clone the repository:

git clone https://github.com/yourusername/sharp-object-detection-alert.git
cd sharp-object-detection-alert

Create a virtual environment (optional but recommended):

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

Install the required dependencies:

pip install -r requirements.txt

Set up your .env file:

GOOGLE_DRIVE_VIDEO_LINK = "https://drive.google.com/uc?id=ID_GOOGLE_DRIVE_VIDEO"
ALERT_CONTACT_EMAIL = ""

The project has mandatory variables not mentioned here in the README for security reasons!

Running the Project

To start processing a video and analyze for sharp objects, follow these steps:

Run the main script in the root folder:

python app.py

The script will:

• Download the video from the provided Google Drive link.
• Process the video frame by frame using YOLOv8.
• Send each frame to Azure Computer Vision for analysis.
• Use the OpenAI assistant to interpret the results.
• Send an email alert (if a sharp object is detected), notification (if no sharp object is found), or an error notification (if something goes wrong).

Check your email inbox and spam (junk mail):

You will receive an email from: brunodurantec@gmail.com based on the result of the analysis.

• Alert email: When a sharp object is detected in the video.
• Result notification: When no sharp object is detected.
• Error notification: When an error occurs during the processing.

Tips

• Do you want to vary the model, number of frames and the confidence for selection? Access the file src\infra\yolo_frame_selection.py and update the variables on the top!

---

Contact

If you have any questions or need support, feel free to reach out:

• LinkedIn: https://www.linkedin.com/in/brunodurante/
• Email: brunodurantec@gmail.com