Priyanshee Sharma Email & Phone Number

Bengaluru, Ka, In

Over the past two months, I embarked on an exciting project to develop a state-of-the-art program capable of detecting objects from drone footage in real-time. Here's a snapshot of the journey:- Started by diving deep into model research, exploring options like YOLOv5, YOLOv8, SSD, and Faster R-CNN to find the best fit.- Settled on YOLOv8, captivated by its remarkable performance and accuracy enhancements.- Successfully trained the YOLOv8 model using the VisDrone dataset.- Encountered performance lags, especially with RTSP streams, prompting further optimization.- Introduced multithreading, creating separate threads for capturing and processing frames simultaneously.- Implemented frame dropping to manage and drop excess frames efficiently.- Reduced input image size and used mixed precision (FP16) to accelerate inference.- Trained a custom YOLOv8 model for specific object classes like cars, buses, motorcycles, trains, and trucks.- Even with optimizations, persistent lag with RTSP streams led me to explore new solutions.- Explored creating a virtual camera using FFmpeg but faced challenges with video splitting.- Successfully used OBS to create a virtual camera, enabling seamless object detection.- Integrated this virtual camera with OpenCV, achieving efficient real-time detection.- The OBS virtual camera setup significantly improved performance with RTSP streams.- Achieved zero lag during object detection when using a webcam, even in offline mode.- This journey was a testament to resilience and innovation, transforming challenges into stepping stones toward achieving real-time object detection from drone footage.

- Directed the design and development of a user-friendly website using React, resulting in a significant 40% increase in user engagement. - Enhanced website functionality by creating over 30 custom landing pages to improve user experience and navigation.- Designed and implemented efficient methods for data exchange, successfully reducing loading times to an impressive 2 seconds.- Increased visitor numbers by 25% through the creation of a more intuitive and user-friendly website interface.I am proud to have contributed to NaturalisT Foundation's success by leveraging my skills in web development to deliver impactful results.

Bachelor Of Engineering - Be, Artificial Intelligence And Machine Learning (Minor)

Bachelor Of Engineering - Be, Electronics And Communications Engineering

High School Diploma

Education record