Image Enhancement in Low-Light Conditions As a partial fulfillment for Deep Learning, AI5100, we took up a low-light image enhancement task using a deep learning-based method called Zero-Reference Deep Curve Estimation(Zero-DCE). Federated Semi-supervised Medical Image Classification via Inter-client Relation Matching As a partial fulfillment for Visual Computing, CS6450 ABCs of the Digital Certificates with Public Key Infrastructures (PKI) Comparison of Digital Certificates in the chain of trust of a website, Man-in-the-middle (MITM) Scenarios involving Digital Certificates Openssl Tutorials with RSA, Digital Certificates, Digests, SHA, MD5 Secure file transfer between Alice (student A) and Bob (student B), Alice (Browser), Bob (web server) and Charlie (Root CA) Decrypting TLS and HTTPS using Wireshark++ Decrypt TLS handshake and HTTPS messages between your browser and the web server of Bank X Network Intrusion Detection using Machine Learning Techniques Anamoly Detection using Machine Learning Techiniques like KNN, DT, SVM Network Intrusion Detection using Machine Learning Techniques Anamoly Detection using Machine Learning Techiniques like KNN, DT, SVM Cracking WPA2-PSK and analyzing Security of IITH Wi-Fi Cracking WPA2-PSK Passphrase with a password dictionary using airodmp-ng, aireplay-ng and wireshark. Creating Firewall with Raw Sockets Creating a simple Firewall using Socket Programming, Extending the rule set and its operation on Firewall, Performance examination and improvement Improving the firewall performance, Detecting DDos attacks in the network using Firewall MITM and Multi-channel MITM attacks on WPA2-PSK/Enterprise Snoofing, Deauthentication, PSK Capture, Password Cracking Fuse Studio, Video Automation This project automates video generation process avoiding the need of manual voice recordings and slide presentations. Slides created in Google Slides are used to generate videos by implementing Text-to-Speech algorithms and Video generation tools. Clickbait Detection A textual data was characterised using ten important attributes combined with word embeddings and modelled using RNN and LSTM with attention to classify clickbait headlines in online news. Self Diagnosis, Computer Aided Diagnosis A computer aided diagnosis approach to detect potential diseases based on symptoms. Baseline models were Naive Bayes, 4-layerd deep neural network and Gradient Boosted Tree(XG-Boost). best performing model was implemented in detecting Tuberculosis and Hepatitis. Project Rentspace An android application that acts as a mediator for customers (Customer to Customer approach) to address the need to rent, lease and sell available spaces like rooms, apartments, lands, hotels, conference halls, etc., online. Obstacle Avoiding Car using Ultrasonic Sound Sensors Designed and created an obstacle avoiding self-driving car that uses ultrasonic sound sensors directed motor modules for detection, navigation, and avoidance using Arduino.
