Pimpri Chinchwad Education Trust’s
Pimpri Chinchwad College of
Engineering
Pimpri Chinchwad Education Trust’s
PIMPRI CHINCHWAD COLLEGE OF ENGINEERING
NIGDI, PUNE-44
DEPARTMENT OF ELECTRONICS AND
TELECOMMUNICATION
SIGNALS AND SYSTEM
MINI PROJECT REPORT ON
TITLE - OBJECT DETECTION USING MATLAB
By:
Nupur Kulkarni - 123B1E149
Gauri Roy - 123B1E154
Veena Kulkarni Ma'am Dr. K.S. Kinage
Faculty Signature H.O.D (E&TC)
1
Sr. no Title Page no
1 Abstract 3
2 Introduction 3
3 Objectives 4
4 Methodology 5
5 Code /Result 6
6 Conclusion/Future scope 7
7 Reference 8
CONTENTS
2
Abstract –
This project focuses on designing an object detection system using MATLAB
to identify and classify specific objects within an image or video frame.
Object detection plays a pivotal role in numerous fields such as surveillance,
autonomous vehicles, and robotics, providing the foundational technology to
recognise and react to environmental stimuli. By leveraging MATLAB’s
Image Processing and Computer Vision toolboxes, we developed an object
detection model capable of identifying objects based on specified criteria,
demonstrating the feasibility and utility of MATLAB for computer vision
applications.
Introduction -
Object detection is a critical task in computer vision that involves
identifying and classifying objects within an image. MATLAB provides
robust tools for object detection through its Image Processing and Computer
Vision Toolboxes, which allow for the application of various algorithms and
techniques to process and analyse visual data. This mini project aims to
design an object detection model that can effectively detect and classify
objects in images or real-time video, utilising MATLAB as the programming
environment. This approach offers a blend of simplicity, visualisation, and
algorithmic efficiency, making it an excellent platform for prototyping
object detection models.
3
Objectives -
The primary objectives of this project are:
1. To develop a MATLAB-based object detection model capable of
identifying specific objects within images or videos.
2. To implement and experiment with different detection algorithms
available in MATLAB.
3. To evaluate the performance of the object detection model based on
detection accuracy and processing speed.
4. To explore MATLAB’s potential for real-time object detection
applications.
Theory -
Object detection is a process that involves locating objects within an image
and classifying each object into predefined categories. The theory behind
object detection involves a combination of computer vision techniques and
machine learning algorithms, such as:
• Feature Extraction: Extracting distinguishing features of objects,
which may include edges, textures, colours, or shapes.
• Classification: Using algorithms to categorise objects based on
features.
• Localisation: Identifying and marking the position of objects within
the image or video frame.
Common approaches to object detection include traditional methods like
Haar cascades, HOG (Histogram of Oriented Gradients) combined with SVM
(Support Vector Machines), and machine learning techniques, as well as
modern deep learning methods using Convolutional Neural Networks
(CNNs).
4
Methodology -
Step 1: Data Collection and Preprocessing
To develop an object detector, relevant image or video data is collected,
either from existing datasets or real-time video feeds. The data is
preprocessed using techniques like resizing, normalisation, and filtering to
ensure consistency and enhance detection accuracy.
Step 2: Feature Extraction
Feature extraction is performed to simplify the input data while retaining
essential information. We used the HOG feature descriptor in this project to
analyse image structure and differentiate objects based on their shapes.
Step 3: Training the Object Detector
MATLAB’s Computer Vision Toolbox provides a variety of pre-trained
object detection models, including the YOLO (You Only Look Once) and
RCNN (Region-based Convolutional Neural Network) frameworks. For this
project, we experimented with HOG and SVM for a simpler implementation,
given the focus on understanding the basics.
Step 4: Object Detection
The object detector is applied to images or video frames to identify objects
and generate bounding boxes. The model checks each image segment for
potential object candidates and classifies them based on trained features.
Detected objects are marked with bounding boxes, and classification labels
are displayed.
Step 5: Evaluation
The model's performance is evaluated based on metrics such as accuracy,
precision, and recall, derived from the number of correctly detected objects
relative to total detections and missed detections.
5
MATLAB Code -
Output / Result -
6
Conclusion -
The MATLAB-based object detection model successfully detected and
classified specific objects in images, demonstrating the potential of
MATLAB for building prototype object detection systems. MATLAB’s
toolbox support allowed for quick setup and exploration of different
detection algorithms, making it an ideal platform for experimentation.
Although our model was simple, it proved effective in achieving real-time
object detection in controlled scenarios.
Future Scope -
1. Real-Time Application: Expand the project to support real-time object
detection for video feeds, optimising processing speed.
2. Deep Learning Integration: Integrate deep learning models, such as
YOLOv4 or faster R-CNN, to improve accuracy and support more
complex objects.
3. Advanced Preprocessing: Implement advanced image preprocessing to
enhance model accuracy, especially for low-light or complex
backgrounds.
4. Multi-Object Detection: Enable the model to detect multiple object
types within the same frame, making it more versatile for practical
applications.
7
References -
1. MATLAB Documentation for Computer Vision Toolbox. Available at:
https://www.mathworks.com/products/computer-vision.html
2. Dalal, N., & Triggs, B. (2005). "Histograms of Oriented Gradients for
Human Detection." International Conference on Computer Vision &
Pattern Recognition.
3. MATLAB and Simulink Example Code. "Object Detection using
YOLO v2 Deep Learning."
