Nagisetti Rithihas +91-8522908238
ECE B.Tech+M.Tech(5Y): Vision and Intelligent Systems # riqxcareer@email.com
Indian Institute of Technology,Kharagpur # GitHub Profile
Education
• Indian Institute of Technology, Kharagpur 2020-Present
Dual degree: B.Tech in ECE and M.Tech in Vision and Intelligent Systems
• Sri Chaitanya Junior College,Khammam 2018-2020
Physics, Chemistry, Mathematics
Research Experiance
• Supervised Bout Classification, Prof.Ravi Hegde, IIT Gandhinagar May’23 - June’ 23
Summer Research Internship Gandhinagar, IN
– Collaborated on a research project jointly conducted by IIT Gandhinagar and IIT Madras, focusing on Remote
Sensing-based Combat Sports Analytics
– Developed a rule-based algorithm for the classification of seven primary boxing punches using human pose data
– Achieved an 80% accuracy rate on side-view data with a processing time of 10 milliseconds
– Attained a 99% accuracy rate with a K-NN classifier utilizing 5th order Angular encoding and a 92% accuracy
rate with a Random Forest classifier utilizing 2D motion dynamic descriptors
– Our research paper has been accepted for presentation at the CVCI’24 conference, and I am currently developing
a novel 3D method to improve accuracy
• PCG based Heart Murmur Detection in children, Prof.Goutham Saha, IIT KGP Aug’23 - Present
B.Tech Project Kharagpur, IN
– Conducting a comprehensive literature review on PCG (Phonocardiogram) and related signal processing techniques
– Utilizing pediatric data from the Circor Digiscope Phonocardiogram dataset and creating a stratified sample
dataset for experimentation
– Employed Python and MATLAB for pre-processing, PCG feature extraction, and classification techniques
– Currently exploring unsupervised approaches to achieve fine-grained murmur classification
• Summer Research Project April’22 - Aug’22
Guide: prof.Vikranth Racherla, Mechanical Engineering Department, IIT Kharagpur IIT Kharagpur
– Designed a Mechanical Model and developed a Computer Vision program for an Autonomous Tree-Climbing
Coconut Harvester
– Engineered a power-efficient 4-wheeled ring-like mechanical model tailored for tree-climbing applications
– Utilized Raspberry Pi and Stereo Cameras to design and prototype the robot
– Implemented portable Occupancy Grid-based climbing system generated using stereo imaging
• Camera Control LEMFV, Prof.sandeep saha, IIT KGP Jan’22 - Present
Research Group Kharagpur, IN
– Contributed to the Controls team within a collaborative project involving IIT Kharagpur, IIT Kanpur, and Moscow
Aviation Institute, aimed at designing a Long Endurance Martial Flight Vehicle (LEMFEV)
– Assisted the aerospace team in the image processing of wind tunnel data for testing a wing prototype
– Conducted a comprehensive study on suitable Vision-only navigation and localization methods for military surface
applications, particularly for fixed-wing UAVs
• Night Image Enhancement for Driving Safty, Prof.Sudipa Mukhopadhyay, IIT KGP Nov’22 - Dec’22
Winter Project Kharagpur, IN
– Investigated challenges in state-of-the-art Night Image Enhancement (NIE) techniques, focusing on scenarios with
prominent light sources
– Studied different NIE techniques, including those based on fundamentals, deep learning, and generative approaches
– Implemented several research papers using Python, starting from the ground up
Publications
Efficient Boxing Punch Classification: Fine-Grained Skeleton-based Recognition Made Light.Vipul
Baghel,Nagisetti Rithihas,Babji Srinivasan*, Ravi Sadananda Hegde**,Computer Vision and Computational
Intelligence - CVCI 2024
Internships
• Code and Execute May’23 - June’23
Summer Intern(Remote) Kanpur, IN
– Leveraged Open3D libraries to optimize packing processes, accounting for the physical characteristics of 3D objects
– Developed Python programs encompassing tasks like Point Cloud Downsampling, Mesh Reconstruction, Orienta-
tion Correction, and detecting Sharp and Fragile points
– Applied Statistical techniques, including Principal Component Analysis, Gradient Descent, and 3D Segmentation
concepts, to create highly efficient algorithms
– Demonstrated the seamless automation of packing operations, consistently delivering satisfactory results across
various objects
• Productize Technology May’23 - June’23
Computer Vision Intern(Remote) Mumbai, IN
– Explored Depth Disparity in Panoramic Outdoor Scene Capture, addressing practical computer vision challenges
– Conducted a comprehensive Literature Review on 3D Reconstruction Techniques like NeRF and SfM
– Proficiently Utilized 3D Reconstruction Software Tools such as InstantNeRF, NeRFStudio, and Meshroom to
create detailed 3D models from images and point clouds
– Effectively Addressed Feature Correspondence Challenges in Panoramic Capture and fine-tuned hardware param-
eters for improved depth capture
Projects
• Electric Wiring Automation May’23
Automated home wiring component detection using deep learning and YOLO V8
– Automated wiring of electrical boards in building construction using fundamental vision and deep learning concepts
– Manually annotated 2000 images using the CVAT tool to create a high-quality dataset for training and validation
– Utilized YOLO V8 for pose estimation, segmentation, detection, and fundamental image processing
– Addressed sparse dataset issues with data augmentation and N-fold cross-validation
• Salient Object Segmentation Dec’22
Fast and Efficient DL Object Outlining for Precision Results
– Developed a Deep Learning pipeline with bounding box annotation for accurate object outlines in Region-of-Intrest
– Utilized U2-Net in the REMBG module for salient object detection and major contour generation
– Designed a user-friendly GUI with keypress functionality for easy program navigation
– Achieved a speedy 1.3-second average execution time for image outline generation and 0.78 Average IoU
• ASR based control for Handicapped Aid Exoskeleton Nov’22 - Dec’22
Hardware Modelling Competition, Azad Hall, IIT Kharagpur
– Collaborated on the development of software for controlling exoskeleton limb movement through speech commands
– Created an audio dataset of native English accents and tested various Wav2Vec models to find the best fit
– Fine-tuned Facebook’s Wav2Vec 2.0 for specific instructions and created a lightweight real-time model using the
Student-Teacher knowledge learning method
– Collaborated closely with the embedded systems team to integrate the software into the hardware
Related Course Work
Pattern Recognition and Machine Learning Fundamentals of Learning Theory Digital Signal Processing
Linear Algebra and Optimization Models Image and Video Processing Probability & Statistics
Neuronal Coding of Sensory Information Computational Neuroscience Digital Communication
Technical Skills and Interests
Areas of Interest: Computer Vision, Machine Learning, Deep Learning, Signal Processing, Statistics
Languages: Python, Matlab, C++(Basics), LateX
Previous Works:vision-based biomechanics, 3D reconstruction & structural analysis, PCG processing
Software: Microsoft Visio, Draw.io, InstantNERF, AlphaPose, Roboflow, CVAT, Anaconda, and CUDA
Others: Demonstrated expertise in conducting thorough literature reviews and implementing research findings into
practical code solutions
Achievements
• SRIP SELECTION Selected for IIT Gandhinagar’s SRIP program out of 36,000 applicants
• JEE All India Rank Achieved All Indian Rank 1969 among 1.8 million candidates in JEE ADVANCED 2020
