My self Sumanth currently working as Software Engineer - AI at Ice Edge Business Solutions - Canada where my key responsibilities is to lead/mentor the formulation of Deep Learning team and recommend strategic directions on using AI. I'm also as a Visiting Faculty at Satyabhama University(India)
I’m a self-taught Artificial Intelligence/Computer Vision enthusiastic (with about 3 year’s experience in industry and research labs) fascinated by the field of DL/ML/CV to analyze data by making use of modern computing power. Earlier I was working as Deep Learning Engineer at Blincam-Japan, which is a wearable camera that takes pictures with a wink of an eye, where I was thrilled to work on analyzing images and designing/deploying/scaling end to end Object Recognition/Detection/Segmentation, Scene Understanding, Face Recognition, Image similarity/ ranking models, content based visual search, sensor time series data analysis models for IoT devices etc. I was also in charge of applying Deep /Machine Learning techniques to develop wink detection logic which can differentiate between intentional and unintentional winks.
And, as part of my master’s degree in Robotics I designed Defence Object Detection models for the Defence Research and Development Organization-India, which is a public defense research lab. I also have experience in sensor fusion (Lidar and Radar), SLAM.
Excluding my day work- Healthcare Analytics, Medical Image Analysis, and Wild life conservation has always been a part of my research and I applied various Deep Learning based models for Cancer (Breast, Skin, Colorectal) /Pneumonia Diagnosis from CT-Scan/X-Rays, MRI image reconstruction, Human Activity Tracking, Steps/ Sleep/ Calorie/ Heart rate analysis etc. which helps to provide better healthcare services. I developed Deep Learning based models for recognizing and monitoring Indian endangered species from camera traps, which helps the zoologists to study and protect endangered species
Use my skills in Artificial Intelligence/Deep Learning/Machine Learning/Computer Vision to build learning systems that automate the scientific discovery process and apply them for decision making in Robotics, Image Analysis, Automotive, Medical, Business and other real-world environments.
I'm mainly interested in using the latest technologies for social good (like cancer diagnosis, elderly care, healthcare, nature/wildlife conservation etc)
* Lead the formulation of the Artificial Intelligence/Deep Learning team. * Use specific AI skills and experience to solve key business issues (Predictive Lead Scoring, Product Knowledge Graphs / Recommendation Systems, Sales Forecasting etc) * Recommend strategic directions where Artificial Intelligence could benefit the business. * Mentor new team members.
• Developed and tested Deep Learning based Object/ Face Recognition and Scene Understanding models • Developed Deep Learning based Image Ranking/ Similarity models • Applied Deep /Machine Learning techniques and developed wink detection logic to differentiate between intentional and unintentional winks. • Extract and analyse image meta data to add value to the company business initiative
✔Evaluate the performance of students and help them to organise their studies and also provide motivation to them during the course of the program so that they can finish all projects in time. ✔ Supply of resources and advice to better understand the content of the course. ✔ Guidance in the development of the projects when students require it.
• Developed Object Detection models for Defence, to be used in Autonomous Vehicles for surveillance applications • Develop and test reliable Deep Learning models for Autonomous Driving
2016-07 - 2017-05
Pneumonia Detection in Chest X-rays
Train deep learning models to detect Pneumonia in chest X-rays of different patients
2018-3 - 2018-4
Cancer Diagnosis from Colo-rectal Histology Images
Train Deep Learning models to diagnose cancer by training it on Colo-Rectal Histology images
2018-1 - 2018-2
Deep Learning For Skin Cancer Diagnosis
Train Deep Learning algorithms to diagnose skin cancer from the images of moles
2017-11 - 2017-12
Deep Learning For Breast Cancer Diagnosis
Train Deep Learning Algorithms on Digital mammography images to diagnose Breast Cancer
2017-9 - 2017-10
Lesion Segmentation Using Deep Learning
Train Deep Learning models for cancer diagnosis by segmenting the Lesions/tumors in CT-Scan images
2017-7 - 2017-8
Extended Kalman Filters To Fuse The Sensor Data To Detect and Track the Bicycle's Position, Velocity.
Use extended kalman filters and sensor fusion(LIDAR & RADAR data) to detect and track the velocity and position of a bicycle that is moving near our vehicle.
2017-3 - 2017-4
Lifestyle/HealthCare Analytics Using Machine/Deep Learning
* Developed Deep/Machine Learning based Human Activity Tracking, Steps/Sleep/Calorie/Heartrate Analysis model on sensor data from smart gadgets, fitness tracker etc
* Analyse Electronic Health Records to compare quality of hospitals, predicting discharge statuses in emergency wards
2017-3 - 2017-4
Advanced Lane Finding
The goal of the project is to write a software pipeline to identify the lane boundaries and segment the drivable area in a video from a front-facing camera on a car. It was developed using OpenCV and python. This algorithm helps the self driving cars to identify the drivable area and drive safely.
2017-2 - 2017-2
Vehicle detection and tracking
Detecting the vehicles and tracking them is very important for any self driving car, for safe driving. Here we address these problems by developing a OpenCV and Python based software pipeline to detect and track vehicles.
2017-2 - 2017-2
Use Deep Learning to Clone Driving Behaviour i.e here we teach the car how to drive on its own by showing it how to drive. And the learned network gives the predicted steering angles based on the video from the front facing cameras.
2017-1 - 2017-1
Traffic Sign Classifier
Design and develop various deep learning based traffic sign recognition models using Convolution Neural Networks
2016-12 - 2016-12
Finding lane lines on the road
It is an OpenCV and Python based algorithm for finding lane lines on the road, which helps the self driving cars to drive on its own.
2016-11 - 2016-11
Monitoring Indian Endangered Species Using Deep Learning
* Every year there are many species that are becoming extinct or classified as endangered. Wildlife conservation enthusiasts are studying these species to better understand them by placing camera traps in forests. However it is very difficult for Zoologists to analyse such data since they are millions of images captured by camera traps. *This projects helps Zoologists to analyze/monitor few of the Indian Endangered species by using Deep Learning models.
2016-9 - 2017-4
Defence Object Detection and Recognition using Deep Learning
o Created new defence object detection data-set o Developed deep learning based object detection models
2016-7 - 2017-5
Fine Grained Image Recognition
o Created two fine grained image recognition datasets o Design CNN architectures for fine grained image classification
I founded NewTechVeda with a vision to minimize skill deficiency in India by encouraging students/ faculty/ employees to learn and adopt to new technologies which are going to rule the industry and academics. It mainly trains the students to do cutting edge research on various new technologies so that they can excel in their careers.
I'm very much interested in teaching and I conduct workshops and faculty development programs at various universities in the fields of Deep Learning, Computer Vision, Machine Learning etc. Recently he has conducted workshops at University Of Calicut, Loyola College (Chennai), PSG (Coimbatore), Kongu College (Erode) , VIT (Chennai/Vellore) etc.
2016-8 - 2018-11
Self Driving Cars Nano Degree
Gold Medalist (Masters in Robotics). Received it from Indian Finance/Defence Minister