Projects - cedlearn

CEDLEARN - Projects

Automated customer satisfaction recording system based on facial expression using Deep Neural Networks

Customer facial expressions are frequently classified as either negative or positive. Customers who exhibit negative emotion toward a given product are more likely to reject the product, whereas customers who exhibit good emotion are more likely to purchase the product. To recognize client spontaneous facial expressions, a conventional neural network (CNN) is used. The purpose of this project is to recognize consumer spontaneous expressions while the client is seeing certain products. We created a customer support system that uses a CNN that has been trained to detect three sorts of face expressions: joyful, sad, and neutral.

Voice based person identification for security control using Natural Language Processing

Voice recognition system is a system that recognizes and authenticates a user's voice by extracting different features from their voice samples. Speech identification is accomplished by translating the human voice into digital data. The digitized audio samples are then subjected to a feature extraction technique to extract Mel Frequency Cepstral Coefficients characteristics. This research focuses on a safe system that employs speech recognition for a natural language by merging digital and mathematical knowledge utilizing MFCC to extract and match characteristics to increase accuracy and performance.

Recommendation engine for e-commerce based on customer past experience using Deep neural networks

A recommender system has become essential. Netflix, YouTube, Amazon, and so many other businesses use this to the optimum. In this project we are integrating deep neural networks on recommendation systems by using Neural Collaborative Filtering (NCF) and build a system to give accurate results on the ecommerce website.

Gesture Control based patient support system

Patient gesture recognition is an exciting new way to learn about and help patients. Healthcare monitoring systems that are coupled with the Internet of Things (IoT) paradigm to perform remote input acquisition solutions. Wearable sensors and information and communication technology have aided in remote monitoring and recommendations in smart healthcare in recent years. Using series learning, the gesture is detected by analyzing the intermediate and structural elements. The suggested gesture recognition system has the capability of monitoring patient behaviors and distinguishing gestures from routine motions in order to increase convergence.

Computer Vision based automatic access control and attendance system

Face recognition access control allows employees to enter the office quickly and easily while also avoiding fraud. Face ID is now used to unlock our smartphones and other smart gadgets on a daily basis. It's a safe and convenient way to get access to our data. Face recognition is now providing the same level of security and convenience to our physical settings. Due to a shift toward health-conscious solutions that simultaneously offer great security, face recognition access control has swiftly become the rising technology in physical access. The FaceNet algorithm is a powerful and current algorithm that we implement in this project.

PDF text annotation and summary extraction using OCR technique

Text in photos contains vital information for indexing and retrieval, as well as automatic annotation and image structuring. As a result, text extraction is an important part of the picture analysis process. Because of the differences in text size, font, style, orientation, and alignment, as well as the complex background, text extraction is a difficult operation. Several text extraction approaches have been developed, including edge detection, linked component analysis, morphological operators, wavelet transform, texture characteristics, and neural networks. This project entails extracting text using OCR and then summarizing the retrieved material according to the requirements.

Automated customer sentiment analysis and risk prediction using Deep Neural Networks

Sentiment Analysis is defined as the process of identifying and extracting information from text in order to better comprehend a brand's social sentiment. It's a text categorization program that analyses incoming messages and displays positive, negative, or neutral attitudes. The absence of appropriate labelled data in the field of Natural Language Processing is a barrier for sentiment analysis (NLP). To address this problem, sentiment analysis and deep learning approaches have been combined, as deep learning models are more successful due to their capacity to learn on their own. Deep Learning makes it possible to process data in a much more complex way. The LSTM, or Long Short-Term Memory model, is a form of Recurrent Neural Network (RNN) that is used to handle temporal data. We employ this neural network design because we believe the order of characteristics (words) in a sentence is important.

Automatic vehicle access control and billing at toll-gate using CNN Algorithm

In Intelligent Transportation Services (ITS), automatic license plate recognition (ALPR) and toll gate billing systems are critical for effective law enforcement and security. Electronic Toll Collection, the new era of intelligent transportation systems, enables the automatic collection of Toll fees from the prepaid account via RFID, thanks to the revolution in communication and embedded technologies. Despite the existing system's lack of security, number plate detection and identification can be done. The current system uses a condition random field technique to detect anomalies. Optical character recognition is used to implement the recognition process. A system may be created to automatically extract the number plate from a car using image processing techniques, match it with a database, generate the One Time Password (OTP) and bill without delay, and identify stolen vehicles. This is accomplished through the use of image processing and motion capture technology.

Enabling blind person to read the text from the pictures through auto voice system using Deep Neural Networks

Using an auto voice-based system to enable visually impaired persons to read text from photos entails extracting the text from the image using OCR and then preprocessing and converting the required text into a voice-based text so that the relevant information can be supplied to the blind people. To assist blind individuals, this procedure involves extracting text from an image and turning the acquired text into voice (text to speech).

AI-based Driver Drowsiness detection and alert system

Drowsiness, defined as a state of drowsiness that occurs when one needs to rest, can result in symptoms that have a significant impact on work performance. The photos of the driver will be examined using artificial intelligence (AI) techniques like deep learning to determine whether or not the driver is drowsy. The technology will be able to alert the driver and prevent accidents utilizing this information. To preprocess the photos of the driver, a combination of artificial intelligence algorithms and deep learning is used. The face will be identified using a linear supporting vector machine (SVM) mixed with a histogram of oriented gradients (HOG), and the driver's landmarks will be detected using an ensemble of regression trees. The preprocessing will continue once the driver's face has been found, and a CNN with the cropped face as an input will be used to determine whether the driver is yawning or not. The parameters collected during the preprocessing phase are then entered into a fuzzy inference system using fuzzy logic, where they will be examined to determine the driver's level of drowsiness.

AI-based collison avoidance system for autonomous cars

Most automakers, including Volvo, Audi, Mercedes, and, most recently, Ford, have released numerous versions of their self-driving vehicles, with some still in the research phase of developing a collision avoidance system. Collision avoidance systems are implemented in a variety of ways in the industry, and the type of sensors, programming models, and hardware employed in the system all play a role. Deep learning is being used to solve the problem of autonomous collision avoidance for a miniature robotic car. Transfer learning technique VGG16 deep network is utilized to accomplish this. The deep network has been employed in real-time with the robotic car. The robotic automobile transmits images to a remote computer that manages the network. The network's predictions are given back to the autonomous car, which then takes action based on them. The findings demonstrate that deep learning has a lot of potential when it comes to resolving the collision avoidance problem.

Early Fire detection system using deep learning and OpenCV

In today's surveillance environments, the technologies that underpin fire and smoke detection systems are critical to assuring and delivering optimal performance. In fact, fire can result in substantial loss of life and property. a deep learning method that leverages a convolutional neural network is used. We evaluated our method by training and testing it using a custom-built dataset of fire and smoke photographs that we acquired from the internet and manually categorized.

Identification of Pathological Disease in Plants using deep learning

Crop diseases are a huge danger to food security, but due to a lack of infrastructure in many regions of the world, timely detection is challenging. Smartphone-assisted disease detection is now achievable thanks to a combination of rising global smartphone usage and recent advancements in computer vision made possible by deep learning. We trained a deep convolutional neural network to identify 14 crop species and 26 diseases using a public dataset of 54,306 photos of damaged and healthy plant leaves taken under controlled settings.

Application to predict election results by performing sentiment analysis on twitter data

The computational study of opinions, sentiments, assessments, attitudes, viewpoints, and emotions represented in text is known as sentiment analysis. It's a classification task in which the goal is to predict the polarity of words and then categorize them as positive or negative. Sentiment analysis on Twitter provides users with a quick and easy tool to gauge popular sentiment toward their party and leaders. We used VADER Sentiment Analysis to perform sentiment analysis on tweets for this project.

Application to detect age and gender of a person using live camera

The recognition of a person's age and gender via a live camera is an important technology that could help salespeople better understand their consumers in any web contact. We'll use MTCNN, one of the most advanced facial recognition models, to recognize faces on the webcam in this implementation. These photographs are fed into deep learning algorithms, which determine the person's age and gender. We'll be creating our own bespoke models to train. However, we have considered transfer learning strategies for less effort and greater accuracy. Many pre-trained models are available, including VGG-face, FaceNet, and GoogLeNet. It's worth noting that the input size requirements for various pre-trained models may differ. As a result, the faces that have been detected must be handled appropriately.

Alternate Route Analysis ( Automated traffic control by providing alternate routes)

Active transportation management solutions are needed to assist agencies in identifying suitable diversion routes for highway incidents, as well as the requirement to change traffic signal timing under various incident and traffic conditions. Using incident attributes and traffic status on the freeway, as well as travel time on both the freeway and alternative routes during the incident, this project investigates the use of a data analytic approach based on the long short-term memory (LSTM) deep neural network method to predict alternative routes dynamically. Additionally, based on the results of the LSTM neural network, simulation modelling, and signal timing optimization, a methodology for developing customized signal plans for crucial crossings on alternate arterials is provided.

Image Based Cancer Cells detection using Deep learning network (CV)

A wide spectrum of biomedical research and therapeutic practices are interested in the ability to automatically detect specific types of cells or cellular subunits in microscope pictures. Methods for detecting cells have progressed from hand-crafted characteristics to deep learning-based algorithms. The basic notion behind these approaches is that their cell classifiers or detectors are trained using labelled target cell locations. In this study, we present a cell detection method based on convolutional neural networks (CNNs), as well as transfer learning techniques for improved accuracy.

Image Based Identify the leaf or fruit ripeness through colour identification(CV)

Agriculture has benefited from the use of image processing in areas such as yield estimation, disease detection, fruit sorting, irrigation, and maturity grading. Image processing techniques can be utilized to save time and money. Because manual sorting does not always produce satisfactory results, it is necessary to employ effective smart farming strategies that can produce higher yields and growth with fewer human resources. Friut detection and analysis is done in this project using image processing techniques (CNN).

Welcome to CEDLEARN, a world of customized training programs in the field of emerging technologies. We have a gamut of programs covering Python Language, Data Analytics, Data Science, Artificial Intelligence, and Full-Stack Development.

What Exactly is Machine Learning?

In simple form, machine learning is all about teaching the machines to learn from the data without explicitly programming. It means machines learn from data i.e. experience just the way we humans learn. Until now we thought that learning is a unique characteristic of living things. As part of machine learning you would use various algorithms and statistical models. Python programming language is widely used here. Machine learning is further classified into 3 types – Supervised, Unsupervised & Reinforcement learning. It is used in various domains such as Banking, Marketing, E-commerce, Education, Medical and so on, wherever there is data and you are interested in extracting useful insights from it for business decisions. Fraud detection, Customer churning, sales prediction, Cancel cell detection etc are a few examples of machine learning in the real life.

What is a Data Science Course?

Data Science is a combination of Data Analytics and Machine Learning. It is all about working on the data to extract the insights from it so that the businesses could make informed decisions. Data Science course would give you the knowledge and skills required to extract the data, clean & manipulate it, use visualization techniques, develop statistical & machine learning models, etc to be the sought after professional. Imagine you are a manager at UBER. You are expected to make decisions related to the number of cars on the road, peak and non-peak pricing, predict the demand during holidays, customer churning, driver analytics and so on. It is humanly impossible to make decisions blindly without having Data Science models. This is the reason why every industry is heavily investing in data science models to be more profitable and competitive.

What is Data Analytics?

Data Analytics is all about extract information from the available data for decision making. It involves Statistical Analysis, Visualization, Dashboarding, Business Intelligence, etc. It mostly deals with tools such as Power BI, Advanced Excel, IBM Cognos, etc. There no or minimum programming involves. Sometimes you would find Python as part of the Data Analytics as it could be used for visualization and analysis. This field is in high demand as not every organization is interested in building whereas, everyone is interested in analysing the data for decision making. Hence, we highly recommend this course for students even those who are from non-technical field. It is good starter to enter in to the world of Data Science where you could test your skills and proceed further.

What Qualifications do I Need to Study Data Science?

You do not need any specific qualification to know Data Science. For that matter, it is true for any course in IT. However, you must have logical thinking, few math skills, algorithmic thinking, loads of patience and a zeal to learn. Most of these requirements could be fulfilled if you are really interested. Of course you need a good trainer or a mentor to handhold throughout your journey towards Data Science. You could start with brushing up on Statistics and learning Python. Commit to the course which is based on project-based learning as mere knowledge would not help you. Practice is the key to being good at Programming or even Data Science. That is the reasons why we give a dashboard for every learner so that they could practice their skills regularly.

Is Data Science a Good Career?

Indeed it is. If you look at the amount of data we are creating on a daily basis while using social media, e-commerce, transportation, online search and even at a retail store. Organizations are interested in processing these volumes of data to get insights so that they could make business decisions such as offering personalized services to the customers or aligning their prices or launching new products or such. This is the very reason why organizations are investing heavily in Data Science models. With this, there is increasing demand for professionals with Data Analytics or Data Science certification to be part of their team and solve business problems. As per the current market trends, Data Science would create 3 times more jobs in the next decade. Hence, it is good timing to think about Data Science as a career option.

What is Hybrid Learning?

Hybrid learning is a revolutionary teaching methodology that encapsulates experiential learning through mentor-led online 1-on-1 session along with video-based e-learning for reinforcement and to suit the learning style of every learner. Once a student enrols (s)he would get a personalized dashboard with detailed learning schedule. Learner would also get access to the self-learning short videos that are sorted as per the schedule so that the learner would watch these short videos to get understanding about the topic before attending online session. This way learner could engage with trainer at a high-level. We believe in project-based learning where the trainer would engage the learner in problem solving to ensure maximum transfer of knowledge and confidence building.

Do you offer a Job Guarantee program?

There are two types of courses we offer one with job readiness and the other with a job guarantee. Even though the course content, projects and approach are the same for both the programs, the only difference lies in the post completion of the course. Irrespective of what type of course you pick, with or without a job guarantee we would train you with profile building, assessments, interview skills, mock interviews and so on. We have partnership with various organizations and also our students are currently working with various reputed organizations. In the job guarantee program we would ensure and work with the learners until they are placed. You would be assigned a career coordinator who would make sure that you are placed or you would get a refund. Currently the packages for a good Data Analyst or Data Scientist could start anywhere between 5L PA to 9L PA at the entry level.