Innovative Machine Learning Projects

Table of Contents

This article list outs the innovative machine learning projects for, diploma & engineering students & researchers. Machine Learning is a branch of Artificial Intelligence, which will able to analyze the data and predict the outcome. In this article, you can go through the computer vision based machine learning algorithms i.e., SSD, YOLO, Haar Cascade. 

Usage of SSD (Single-Shot Multi Box Detector) and YOLO (You Only Look Once) is a convolutional approach of identifying multiple objects in an image/video. Basically these algorithms will able to classify the image and identify the object.

The Haar Cascade is a machine learning based approach where a lot of positive and negative images are used to train the classifier. The positive images contains the things that the classifier wants to identify and the negative images contains the things that the classifier shouldn’t have to identify.

In order to design an application with these algorithms to work as a device, we choose Raspberry Pi Processor. All of the Innovative machine learning projects in this article requires a common library i.e., OpenCV. And for conversion of text to speech, we need pyttsx3.

Top 5 Innovative Machine Learning Projects

1. Object Detection with Voice Over

The technology is getting advanced day by day in a way that a machine can able to identify the object and inform you with a voice over. You can use this concept in an application where it helps the blind person by letting him know what object is present in front of him.

The object detection is a field of computer vision which will identify the semantic objects in a video/image. The most popular algorithms that we use are SSD, YOLO. We can then convert the identified object name which is in the text format into a voice response. This is how one can able to use this advanced technology to guide themselves.

2. ML based Surveillance system for detection of Triple Riders

As the use of vehicles are increasing daily, the occurrence of road accidents were also been increasing which will lead to death of a human. Many of the people who died in such an accidents is due to not wearing helmets and riding as triples. This system will identify those riders and gives an alert to authorities. So that such accidents will be diminished.

In this project we have interfaced a wireless communication between the MATLAB and the embedded system. The MATLAB will do the process of identifying the riders without helmets and triple riders using machine learning. If any rider was found like that, a unique data will be sent to the embedded system. In our embedded system, we are having a Raspberry Pi processor which will receive that unique data. 

Once the data was read, the embedded device will able to get the vehicle details with the help of an RFID tag that was attached to the vehicle. So thus the extracted data, from the tag which contains the vehicle number along with that the data received from MATLAB which will tell us whether the rider is not wearing helmet or triple riders, will be sent to authorities in order to take an action.

3. Smart Attendance System

Generally where ever you see, in colleges/schools taking attendance of students by teachers and taking attendance of teachers by management; in industries/organization taking attendance of an employee, is all a time taking process which will show its effect on daily productivity. is also As the name itself is a smart attendance system, the concept that advanced which will able to recognize the person face and takes the attendance.

Using OpenCV, we are able to detect the face and recognize the face. For detecting the face, we have to use Haar Cascade algorithm. And for recognizing the face, we have to use LBPH (Local Binary Pattern Histogram) algorithm. In order to recognize a person face, dataset of his face should be there in the processor (Raspberry Pi). 

A dataset should contain all the faces of the people in a school/college/organization/industry with individual unique ids. Once recognized, the Raspberry Pi will able know that unique id of the identified person and will upload the data to an open source cloud platform.

The cloud platform will contain the data of unique id of an identified person and the time when he/she entered in to the classroom/organization.

4. Covid-19 Face Mask Detection

In order to reduce the rapid spread Covid-19, WHO has given some prevention measures of corona virus. Out of that one is to wear mask in public places which will reduce the spread of virus while coughing, sneezing. Though many people neglected in the first place, as the effect of covid is getting higher the public places like shopping malls, hospitals, cinema theatres people were taken some action to not allow anyone who is not wearing mask. 

Manually a person couldn’t stand outside for hours and find whether anyone is not wearing mask and what is their body temperature. So we built a system which will identify the person body temperature and whether he is wearing mask or not.

For identifying whether a person is wearing mask or not, we have to use Opencv Haar Cascade algorithms.  And for identifying the body temperature, we have used an IR based body temperature sensor. If person who is entering should have a normal body temperature and also have to wear mask. 

If he/she is not having a normal body temperature or not wearing mask, the door won’t open. So that we can reduce spread of corona in public places by making sure everyone is wearing mask and having normal body temperature.

5. Smart EVM with Raspberry Pi

Earlier we used Paper ballot for voting, later on improvisations are happening and now Electronic Voting Machine. In this concept, Camera is used to detect faces. Initially camera captures set of images and stored in Dataset folder. So voter1 dataset is created and similarly corresponding voters datasets are created. After that, you need to train all the faces that are stored in Dataset. 

During elections, while voter entering into the voting hall, camera detects the face and ask to enter the key for voting. The voter will vote for respective party using Keypad and immediately this data will be uploaded to the server. This is how you can develop a smart Electronic Voting Machine with Raspberry Pi using Facial Recognition.


So here are few innovative machine learning projects. You have gone through different algorithms based applications such as object detection based on SSD, YOLO and Haar Cascade algorithm based face recognition and face mask detection. And text to speech conversion based application. 

In order to gain practical knowledge and Hands on experience with Machine Learning based Raspberry Pi projects, go through

Final year projects