real-time-object-detection

pred_video_compressed2.mp4

Table of Contents

1. Installation Instructions
   • Prerequisites
   • Dependencies
2. Usage
   • Running the Application
   • Example
3. Features
4. Configuration

Installation Instructions

Prerequisites

Before you begin, ensure you have met the following requirements:

• You have installed Python 3.6 or later.
• You have a working installation of Jupyter Notebook.
• You have pip installed.

Dependencies

This project uses the following libraries and packages:

• argparse
• os
• matplotlib
• scipy
• numpy
• pandas
• PIL
• tensorflow
• yad2k

You can install the required packages using the following command:

pip install argparse os matplotlib scipy numpy pandas pillow tensorflow git+https://github.com/allanzelener/YAD2K.git

Usage

Running the Application

To use this project, follow these steps:

1. Clone the repository:

   git clone https://github.com/justinliu23/real-time-object-detection.git
   cd <repository_directory>

2. Launch Jupyter Notebook:

   jupyter notebook

3. Open the notebook file Autonomous_driving_application_Car_detection.ipynb and run the cells sequentially.

Example

The application processes images to detect cars. Here is an example of how to use the provided functions:

1. Load and preprocess an image:

   image, image_data = preprocess_image("path_to_image", model_image_size=(608, 608))

2. Load the YOLO model:

   yolo_model = load_model("model_data/yolo.h5")

3. Run the model and get the output:

   yolo_outputs = yolo_head(yolo_model.output, anchors, len(class_names))

4. Filter boxes:

   scores, boxes, classes = yolo_eval(yolo_outputs, image_shape)

5. Draw bounding boxes:

   draw_boxes(image, boxes, classes, class_names, scores)

Features

• YOLO Model Integration: Utilizes the YOLO (You Only Look Once) model for object detection.
• Image Preprocessing: Supports image preprocessing including resizing and normalization.
• Bounding Box Visualization: Visualizes detected objects with bounding boxes.
• Confidence Thresholding: Filters detections based on confidence scores.

Configuration

Configuration details for the YOLO model and detection parameters are as follows:

• YOLO Configuration Files: The project uses pre-trained weights and configuration files available from the YAD2K repository.
• Non-Max Suppression Threshold Settings: Modify the class score threshold as needed. For example:

  threshold = 0.6