Real-Time Object Detection with TensorFlow and OpenCV

This repository contains Python scripts for implementing object detection using TensorFlow and OpenCV. The project uses a pre-trained SSD Mobilenet V2 model from TensorFlow Hub for detecting objects in real-time from either image or video files or webcam streams.

Features

• Pre-trained Model: Utilizes TensorFlow Hub's SSD Mobilenet V2 for efficient object detection.
• Real-Time Processing: Supports image(real3.py) as well as video files (real.py) and live webcam streams (real2.py).
• Visualization: Draws bounding boxes with class names and confidence scores on detected objects.

Requirements

• Python 3.7 or later
• TensorFlow
• TensorFlow Hub
• OpenCV
• NumPy
• ScreenInfo

Installation

Step 1: Clone the Repository

Use the following commands to clone this repository and navigate to the project directory:

git clone https://github.com/SnehalSanap0/Trace.git
cd Trace

Step 2: Install Dependencies

Install the necessary dependencies using the requirements.txt file. Run this command:

pip install tensorflow tensorflow-hub opencv-python numpy screeninfo

Step 3: Run the Scripts

Once the dependencies are installed, you can execute the scripts:

To process video files (real.py):

python real.py

To detect objects using your webcam (real2.py):

python real2.py

To process image files (real3.py):

python real3.py

Usage

Running real.py Place your video file (e.g., classroom.mp4) in the project directory. Run the script:

python real.py

The output window will display detected objects with bounding boxes.

Running real2.py Connect your webcam. Run the script:

python real2.py

Press q to close the detection window.

Running real3.py Place your image file (e.g., image3.jpg) in the project directory. Update the image file path in the script:

image_path = "image3.jpg"

Run the script:

python real3.py

The script will display the image with detected objects highlighted. Press any key to close the window.

Customization

Video Input for real.py: Update the video file path in the script real.py to use a different file:

video = cv2.VideoCapture("your_video.mp4")

Resolution for real2.py: Modify the resolution settings in real2.py:

video.set(cv2.CAP_PROP_FRAME_WIDTH, 640)
video.set(cv2.CAP_PROP_FRAME_HEIGHT, 480)

Image Path: Update the image_path variable to specify the file you want to process:

image_path = "your_image.jpg"

Detection Threshold: Change the score_threshold in the draw function to filter detections by confidence level:

score_threshold=0.1  # Increase to show fewer, higher-confidence detections

Troubleshooting

TensorFlow Hub Loading Issues:

Ensure your internet connection is active when running the script for the first time.

Webcam Not Working:

Check that the correct webcam index (0 for default camera) is used in real2.py:

video = cv2.VideoCapture(0)

Performance Issues

Reduce video resolution to improve performance in real2.py. Use smaller input images to speed up processing. Adjust the detection threshold to reduce the number of displayed objects.

Contributing

Contributions are welcome! Feel free to open an issue or submit a pull request to improve the project.