Saptarshi Mandal

About Me

I am a Signal Processing and Machine Learning Student.

I am a Masters Student at Indian Institute of Science, Bangalore working in the Spectrum Lab under the guidance of Prof. Chandra Sekhar Seelamantula.

My research interests are Image Processing, Computer Vision and Signal Processing.

I primarily code in Python and am proficient in Deep Learning Frameworks such as PyTorch and Tensorflow. I also have published research articles in top peer-reviwed journals.
I have had the priviledge to work at the following Institutions/Organisations:

Education

2023 - 2025

Master of Technology (M.Tech)

Signal Processing

Indian Institute of Science

Bangalore, India

CGPA : 9.30 / 10.0

2018 - 2022

Bachelor of Engineering (B.E.)

Electrical Engineering

Jadavpur University

Kolkata, India

CGPA : 9.12 / 10.0

2016 - 2018

Secondary Education (Class 12)

Science (PCMB)

Nava Nalanda High School

Kolkata, India

Percentage : 91.4%

Experience

2024 May - 2024 July

Interim Engineering Intern

Qualcomm India Pvt. Ltd.

Hyderabad, India

Team : QCT Camera Systems
Project Title : Enhancement on AI Camera Session Analyzer
Project Objective : Developing an ADB and DSP Log Parsing Tool for AI Camera Execute Sessions
2022 July - 2023 June

Management Trainee

CESC Ltd.

Kolkata, India

Deptartment : Mains
Tasks : 6kV/11kV Fault Management, LT Cable Fault Management.
2021 June - 2021 July

Intern

CESC Ltd.

Kolkata, India

Deptartment : Construction
Project Title : Study of Voltage Disturbances and Protection Co-ordination in Stations LTAC Network.

Projects

Beyond the Canvas of One: Expanding the Horizons of Single-Image Generation

March 2024 - May 2024

Associated with Indian Institute of Science, Bangalore

This project aims to enhance the SinGAN model (Shaham et al., ICCV, 2019) and overcome its limitations, particularly its reliance on a single input image and sensitivity to image quality. We propose to develop a more robust and versatile generative model
Exploring the Impact of Image Pre-Processing on Human Face Detection in Poor Quality Images

October 2023 - November 2024

Associated with Indian Institute of Science, Bangalore
- Implementation of Wiener Filtering, PCA, Non-Local Means based Image Reconstruction, and Haar-Cascade Detector for Face Detection.
- Achieved IoU increase of 0.34 and Average Recall increase of 0.46 after Image Pre-Processing
Time Frequency Analysis of Smartphone Sensor Data for User Identification

August 2020 - March 2021

Associated with Jadavpur University, Kolkata
- Feature Extraction using FFT and Short-Time Fourier Transform (STFT)
- Proposed CNN Model demonstrates superior performance (Accuracy = 0.98, Precision = 0.96, Recall = 0.98 and AUC = 0.995) compared to LeNet, AlexNet and ZFNet for Gait Based User Identification