Hi I am Pranjal Rai, an undergraduate student at IIT Delhi in the Department of Electrical Engineering, working towards a B.Tech in Electrical Engineering.
My research interests are in Computer vision, Image Processing and Embedded Systems. I have worked on research projects in Flexible electronics, Deep Learning and Embedded Systems.
I have designed and
developed an embedded systems device which captures retinal images and predict ECG by analysing the retinal vasculature. I have also contributed in development of a novel Flexible Piezoelectric tactile sensor. Apart from academic work,
I am also active in social service and concerned about the global problem of Climate Change, to which I think Geo-Engineering could provide a potential technical solution.
1. Rajinder S. Deol, Nitika Batra, Pranjal Rai and Madhusudan Singh,
Electric Field Induced Annealing of Inorganic thin films for Densification, Indian Provisional Patent Application No. 201911049331 filed on November 30, 2019
2. Rajinder S. Deol, Nitika Batra, Pranjal Rai, Henam Sylvia Devi, Bhaskar Mitra and Madhusudan Singh,
Electrically Annealed and Densified Solution-Deposited Piezoelectric Thin Films, 2019 MRS Fall Meeting and Exhibit: Symposium FF04.05.16
"I will rather have questions that cannot be answered than answers that cannot be questioned" - Richard Feynman
PROJECTS
Embedded System Decive to measure ECG using Retinal Analysis (Embedded Systems, Department of Computer Science and Engineering, IIT Delhi)
Currently, I am working under the guidance of Prof. Kolin Paul at the Department of Computer Science and Engineering, IIT Delhi. We are developing an Embedded System device which can be integrated with an ophthalmoscope. Our device uses a Raspberry Pi-3 processor and a Raspberry Pi camera to record videos of the retina through the ophthalmoscope.
The videos are sampled and the resulting images are processed to quantify the amount of Pulsations near the opticla disc by measuring and recording the time variation of diameter of blood vessels. The variations in diameter of the
blood vessels are further processed to predict ECG and detect the presence of major systemic vascular diseases, including ischemia, coronary heart diseases, and diabetes mellitus and its complications.
Fetal head segmentation and circumference measurement (Individual project on Computer vision and Deep learning)
I Implemented and trained a U-Net model with ResNet inspired residual connections for the segmentation of fetal head from ultrasound images. Used the data from the HC18 challenge to train the model. I further Implemented Elliptic Hough transform to fit an ellipse to the segmented images. Achieved an average Dice score of 94% on the test data.
Implementation of Various 5G Modulation Techniques (Signal Processing, Department of Electrical Engineering, IIT Delhi)
I worked with Dr. Gaurav Pandey under the guidance of Prof. Amol Choudhary at the Photonics Lab,
IIT Delhi. We used MATLAB to implement various 4G (LTE and A-LTE) and 5G modulation techniques including OFDM (Orthogonal Frequency Division Multiplexing), SC-FDM (Single Carrier Frequency Division Multiplexing), GFDM ( General Frequency
Division Multiplexing), F-OFDM (Filtered OFDM), UFMC (Universal Filter Multi-Carrier modulation) and FBMC (Filter Bank Multi-Carrier modulation). We have setup an Analog Photoinc Link (APL) and have incorporated a wireless transmission
link after the APL, using dipole-log periodic and horn antennas. We also analyzed the APL for various characterictics like SFDR (Spurious Free Dynamic Range), DR (Dynamic Range), NF (Noise Figure) and G (Link Gain) through experimental
setup. We were analyzing the modulated waveforms after passing them through the APL for various characteristics like EVM (Error Vector Magnitude), BER (Bit Error Rate), PAPR (Peak to Average Power Ratio), etc.
Flexible Piezoelectric Tactile Sensors and Investigation of Electric field Induced Annealing of Inorganic thin films (Nanoelectronics, Department of Electrical Engineering, IIT Delhi)
In the summers of 2019, I worked with Mr. Rajinder Singh Deol under the guidance of Prof. Madhusudan Singh at the FMD Lab, IIT Delhi. We developed Felxible KNN (Potassium Sodium Niobate) based tactile sensors. We used sol-gel based method to synthesize KNN inorder to reduce the thermal budget
of the devices. We optimized the KNN sol layer and the design of the device. We used metal (Al and W) deposited via PVD (Physical Vapour Deposition) technique as the substrate for the devices. We characterized the devices for various
parameters like Pressure Sensitivity, Flexibility, Decay rate etc. Data was collected using Labview program and was analyzed using MATLAB. Our work was presented at the EMC (Electronic Materials Conference) held at the University of Michigan from June 26, 2019 to June 28, 2019. Title of the presentation was "Lead-Free Tactile Sensors and Energy Harvesting Devices for Wearable and Flexible Electronics", Symposium P05. We also investigated the phenomena of Electric Field induced Annealing in thin KNN films. We applied high electric field on thin deposited KNN films. We analyzed the effect of the electric field on the material using SEM images, AFM
measurements and EDX measurements. We used MATLAB to claculate the percentage volume reduction in the material. Our work was presented at the MRS (Materials Research Society) 2019 fall
meeting held at Boston, Massachuetts from December 1, 2019 to December 6, 2019. Title of the presentation was "Electric Field assited annealing of KNN thin films for densification",
Symposium FF04. We also filed a Provisional Patent for the above developed method of annealing. Indian
Patent Application No.:201911049331
Melanoma prediction using ensamble of FCNN and CNN (Individual project on Computer vision and Deep learning)
I Trained an ensamble of EfficientNet-B7 and a 500-250 Fully Connected NN (FCNN) for the prediction of Melanoma from lesion images and patient information. Used the data from the SIIM-ISCI challenge I further Incorporated various image transformations along with painting artificial hairs on the images, to extend the training set. Images along with the tabular data of the patient was used as an input to the model.
Data Structures, Algorithms and other Course Projects (Department of Computer Science and Engineering, IIT Delhi)
I have also worked on various other Programming projects as a part of my Data Structures and Algorithm Course at IIT Delhi. You can find my programming projects
on my page.
