LeadingIndia.AI

LeadingIndia.AI

Data Scientist

June 16, 2026 – Present

ROUTING-ALGORITHMS-FOR-ENERGY-EFFICIENCY-IN-UNDERWATER-WIRELESS-SENSOR-NETWORKS

July 8, 2020 – July 8, 2020

ROUTING-ALGORITHMS-FOR-ENERGY-EFFICIENCY-IN-UNDERWATER-WIRELESS-SENSOR-NETWORKS — GitHub repository

Wake-UP-word-detection

March 11, 2019 – March 11, 2019

Wake-up-word(WUW)system is an emerging development in recent times. Voice interaction with systems have made life ease and aids in multi-tasking. Apple, Google, Microsoft, Amazon have developed a custom wake-word engine, which are addressed by words such as ‘Hey Siri’. ‘Ok Google’, ‘Cortana’, ‘Alexa’. Our project focuses initially only detection and response to a customized wake-up command. The wake-up command used is “GOLUMOLU”. A wake-up-word detection system search for specific word and reads the word, where it rejects all other words, phrases and sounds. WUW system needs only less memory space, low computational cost and high precision. Artificial Neural Networks(ANN) have reduced the complexity, computational time, latency, thus the efficiency of system has improved. Deep learning has improved the efficiency of automatic speech recognition(SR), where wake word detection is a subset of SR but unlike keyword spotting and voice recognition. A deep learning RNN model is used for the t

Click-Fraud-Detection

September 27, 2018 – September 27, 2018

Click Fraud is a type of fraud that occurs on the Internet in pay-per-click (PPC) online advertising. It occurs by intentional clicking of online advertisements with no actual interest in the advertised product or service. Click Fraud is an important threat to advertisement world that affects the revenue and trust of the advertisers also. To tackle this issue, we use various machine learning and deep learning model that can learn from the data given to train the system. Model will identify the fraud clicks based upon the data provided to the system in training stage. We made a deep learning model to classify the fraud and non-fraud clicks and compared its result with various machine learning approaches like SVM, Naive Bayes and logistic Regression.

Forest-Fire-Detection-through-UAV-imagery-using-CNNs

September 13, 2018 – September 13, 2018

Wildfire is a natural disaster, causing irreparable damage to local ecosystem. Sudden and uncontrollable wildfires can be a real threat to residents’ lives. Statistics from National Interagency Fire Center (NIFC) in the USA show that the burned area doubled from 1990 to 2015 in the USA. Recent wildfires in northern California (reported by CNN) have already resulted in more than 40 deaths and 50 missing. More than 200,000 local residents have been evacuated under emergency. The wildfires occur 220,000 times per year globally, the annual burned area is over 6 million hectares. Accurate and early detection of wildfire is therefore of great importance. Fire detection task is crucial for people safety. Several fire detection systems were developed to prevent damages caused by fire. One can find different technical solutions. Most of them are sensors based and are also generally limited to indoors. They detect the presence of particles generated by smoke and fire by ionization, which require

Chatbot-using-Recurrent-Neural-Networks

September 13, 2018 – September 13, 2018

A conversational agent or a chatbot is piece of software which can communicate with human users with the help of natural language processing (NLP). Modelling conversation is a very crucial task in natural language processing and artificial intelligence (AI). Since the discovery of artificial intelligence, creating a good chatbot is one of the field’s hardest and complex challenges. Chatbots can be used for various tasks such as make phone calls, provide reminders etc; in general they have to understand users’ utterances and provide relevant responses for the problem in hand. Previously, methods which were used for constructing chatbot architectures relied on hand-written rules, templates or simple statistical methods. Rising and innovating field of deep learning have replaced previous models with trainable neural network models. The recurrent encoder-decoder model is the dominating model in the field modelling conversations. Multiple variations and features have been presented that hav

Volume-Control-using-Hand-Gestures-Recognition

September 13, 2018 – September 13, 2018

Gesture recognition helps computers to understand human body language. This helps to build a more potent link between humans and machines, rather than just the basic text user interfaces or graphical user interfaces (GUIs). In this project for gesture recognition, the human body's motions are read by computer camera. The computer then makes use of this data as input to handle applications. The objective of this project is to develop an interface which will capture human hand gesture dynamically and will control the volume level. For this, Deep Learning techniques such as Yolo model, Inception Net model+LSTM, 3-D CNN+LSTM and Time Distributed CNN+LSTM have been studied to compare the results of hand detection. The results of Yolo model outperform the other three models. The models were trained using Kaggle and 20% of the videos available in 20 billion jester dataset. After the hand detection in captured frames, the next step is to control the system volume depending on direction of hand

-Fake-News-Detection-

September 12, 2018 – September 12, 2018

Fake news is misinformation or manipulated news that is spread across the social media with an intention to damage a person, agency and organisation. Due to the dissemination of fake news, there is a need for computational methods to detect them. Fake news detection aims to help users to expose varieties of fabricated news. To achieve this goal, first we have taken the datasets which contains both fake and real news and conducted various experiments to organize fake news detector. We used natural processing, machine learning and deep learning techniques to classify the datasets. We yielded a comprehensive audit of detecting fake news by including fake news categorization, existing algorithms from machine learning techniques. In this project, we explored different machine learning models like Naïve Bayes, K nearest neighbors, decision tree, random forest and deep learning networks like Shallow Convolutional Neural Networks (CNN), Deep Convolutional Neural Network (VDCNN), Long Short-Ter

LANE-DETECTION-USING-DEEP-LEARNING

September 12, 2018 – September 12, 2018

Autonomous self-driving is in the trend for implementing it in our real life to remove all the hassles and accidents. Modern-day transport has come a long way but still far away from perfection and all-around safety. Lane Detection is a concept of demarcating lanes on the roads while the vehicle is moving. It has the capability of changing the vehicular movements on road, making them more organized and safe. This leap could provide for driver carelessness and avoid a lot of mishaps on the roads. Ride-hailing services like Uber and Ola can use them to monitor drivers and rate them based on driving skills. We have designed and trained a deep Convolutional Network model from scratch for lane detection since a CNN based model is known to work best for image datasets. We have used BDD100k dataset for training and testing for our model. We have used various metrics values for hyper-parameters tuning and took the ones which gave the best result. The training is done on Supercomputer NVIDIA-DG

Road-damage-detection

September 12, 2018 – September 12, 2018

Keeping roads in a good condition is vital to safe driving. To monitor the degradation of road conditions is one of the important component in transportation maintenance which is labor intensive and requires domain expertise. Automatic detection of road damage is an important task in transportation maintenance for driving safety assurance. The intensity of damage and complexity of the background, makes this process a challenging task. A deep-learning based methodology for damage detection is proposed in this project after being inspired by recent success on applying Deep- learning in Computer Sciences. A dataset of 9,053 images is taken with the help of a low cost smart phone and a quantitative evaluation is conducted, which in turn demonstrates that the superior damage detection performance using deep-learning methods perform extremely well when compared with features extracted with existing hand-craft methods. Using convolutional neural networks to train the damage detection model wi

Real-Time-Multiple-Object-Detection

September 12, 2018 – September 12, 2018

The ability of the computer to locate and identify each object in an image/video is known as object detection. Object detection has many applications in self-driving cars, pedestrian counting, face detection, vehicle detection etc. One of the crucial element of the self-driving car is the detection of various objects on the road like traffic signals, pedestrian’s other vehicles, sign boards etc. In this project, Convolutional Neural Network (CNN) based approach is used for real-time detection of multiple objects on the road. YOLO (You Only Look Once) v2 Deep Learning model is trained on PASCAL VOC dataset. We achieved mAP score of 78 on test dataset after training the model on NVIDIA DGX-1 V100 Super Computer. The trained model is then applied on recorded videos and on live streaming received through web cam.