Sakshi Sinha Email and Phone Number

• Engineered a high-accuracy Plant Disease Identification Model with 70,295 image datasets across 38 classes, reaching 95% CNNaccuracy; combined with a 99.54% accurate Crop Recommendation Model using Random Forest, optimizing agricultural decision-making. • Worked on predictive models for heatwave systems detection using RoBERTa and Random Forest, achieving 70.42% accuracy. • Implemented a lane detection pipeline using Python and OpenCV, integrating color selection, Canny edge detection, Gaussian smoothing and Hough transform to detect and extrapolate lane lines in images and video streams.

As a Software Engineering intern at INNODEED, my project involved developing a module to determine whether a person's eyes are closed or open in photographs. During this project, I utilized the Mediapipe model to extract facial landmarks, particularly focusing on eye position detection. Leveraging this data, I implemented a ResNet-inspired sequence model for predicting the eye state. To make the module easily accessible and usable, I integrated it with Flask, a web application framework in Python.The goal of this module was to identify and discard bad photographs that may not be suitable for further analysis or processing. By accurately detecting closed or open eyes, the module aimed to enhance the overall quality of images and improve the efficiency of subsequent image-related tasks. This combination of technologies, along with the integration with Flask, enabled accurate eye-state classification and contributed to the development of a robust module that effectively served its purpose in determining if the eyes were open or closed in photographs. Users could easily access and interact with the module through a web interface, making it a practical and user-friendly solution.