Trained a machine learning model to classify hand drawn images. Used MobileNet CNN to achieve an accuracy of 98% with 340 categories.
Developed a machine learning model to detect the intention of any search query (if legal query or not). Preprocessed existing search query data, performed feature engineering using existing resources (google search results, data from law.cornell.edu) and 14 different regexes. Used SVM, Naive Bayes classifiers and LSTM. Achieved an accuracy of around 95%.
MATLAB’s image processing block-based LiDAR
Developed a 3D surround scanner by using MATLAB’s image processing block to scan the surroundings with help of line laser and a camera to plot the 3D graph using MATLAB’s image processing block.
Embedded Systems (Wireless security system)
Programmed in C, the system comprises of RF based remote modules to build the communication layer of the whole system and occupancy sensors to detect intrusion.
Programmed the MCU in C to detect precise movement of accelerometer and print the output on attached screen.