Himanshu Lal

◦ Implemented a Conditional GAN to convert grayscale images to color images utilizing the pix2pix framework
◦ Explored diverse loss functions and generator designs, assessed on ImageNet and MSCOCO via PSNR, FID score & MAE
◦ Constructed a Deep Residual-UNet architecture for the generator, utilizing generator pre-training

◦ Conducted a comprehensive study of climate change factors: GHG emission, Rise in temperature and sea level, Glacier
melting, Pollution and Deforestation, investigating their connections to disasters and their impact on species and environment
◦ Identified the root cause of each climate change factor and predicted potential consequences without major intervention
◦ Conducted in-depth study on climate change factors, identified origins, connections to disasters, and environmental… Show more

◦ Conducted a comprehensive study of climate change factors: GHG emission, Rise in temperature and sea level, Glacier
melting, Pollution and Deforestation, investigating their connections to disasters and their impact on species and environment
◦ Identified the root cause of each climate change factor and predicted potential consequences without major intervention
◦ Conducted in-depth study on climate change factors, identified origins, connections to disasters, and environmental impact
◦ Curated and refined 40+ diverse raw datasets from global data sources and optimized them for comprehensive analysis Show less

◦ Designed an IoT-based Smart Door system, integrated with Face Mask Detection, Hand Sanitization, Temperature
Screening and Tracking Room Headcount, providing a contactless entry-exit experience and health safeguarding solution
◦ Implemented in-build OpenCV’s DNN module for face detection and fine-tuned the MobileNet-V2 model for face mask classification, achieving an impressive 99.62% accuracy on the test set.
◦ Integrated the entire functionality on Arduino Mega and simulated on… Show more

◦ Designed an IoT-based Smart Door system, integrated with Face Mask Detection, Hand Sanitization, Temperature
Screening and Tracking Room Headcount, providing a contactless entry-exit experience and health safeguarding solution
◦ Implemented in-build OpenCV’s DNN module for face detection and fine-tuned the MobileNet-V2 model for face mask classification, achieving an impressive 99.62% accuracy on the test set.
◦ Integrated the entire functionality on Arduino Mega and simulated on Wokwi, for real-world applicability. Show less

◦ Built a smart irrigation system that regulates water flow for edge devices based on real-time humidity and moisture levels
◦ Trained a 2-layer MLP for water flow prediction and simulated its functionality on Arduino 2560 through Wokwi
◦ Over 60% of the prediction model shows 0% deviations, and for over 70% of predictions, deviations are less than 1%

◦ Developed a two-player turn-based game-playing AI utilizing self-play for iterative learning and policy enhancement
◦ Designed a user interface for the chess board using kivy and python-chess
◦ Implemented a CNN-based Monte-Carlo Tree Search algorithm for predicting optimal moves from current states

◦ Developed a Convolution-3D (C3D) network to learn spatiotemporal features from CT-scanned images of lungs
◦ Trained the C3D model on the dataset provided by SPIE, AAPM & NCI, classifying each sample into malignant and benign

•The task was to perform churn analytics on the bank dataset which comes from the Kaggle web site (https://www.kaggle.com/barelydedicated/bank-customer-churn-modeling). The idea was to calculate churn rate by dividing the number of customer’s cancellations within a time period by the number of active customers at the start of that period. The output of a churn model is a measure of the immediate or future risk of customer cancellation. The analytics is done on 10,000 observations and 14… Show more

•The task was to perform churn analytics on the bank dataset which comes from the Kaggle web site (https://www.kaggle.com/barelydedicated/bank-customer-churn-modeling). The idea was to calculate churn rate by dividing the number of customer’s cancellations within a time period by the number of active customers at the start of that period. The output of a churn model is a measure of the immediate or future risk of customer cancellation. The analytics is done on 10,000 observations and 14 features. The idea was to analyze the customers with maximum chances to their cancellations.
•Language and Model used for analytics : Python and Artificial Neural Network
•Tools used : Jupyter Notebook Show less