- STM32 microcontroller to control the robot's thrusters from a desktop computer.
- Raspberry Pi 5 to get a live feed from the robot's camera and display it on a screen.
- Developed an object detection model using YOLO to detect sea life.
A cross-platform system performance benchmarking suite built for in-depth analysis and comparison of CPU, GPU, memory, and cache performance across workloads, architectures, and environments.
- Configurable microbenchmarks: floating point throughput, memory latency/bandwidth, thread scalability
- ML workloads powered by scikit-learn, PyTorch (CPU/GPU/MPS), and TensorFlow
- Compiler benchmarking with gcc / clang via real-world C project compilation
- Detailed hardware info introspection (RAM, CPU cores, frequencies, per-core usage, cache levels)
- Fully JSON-driven configuration for parameterized testing
- Results auto-saved in timestamped .json files for logging and comparison
- Modular and extensible structure using Python packages
- FP Benchmarks: NumPy, Pandas
- ML: PyTorch, TensorFlow, scikit-learn
- DB: sqlite3
- System Metrics: psutil, py-cpuinfo
- pyproject.toml build system + JSON configuration loading
- Cross-platform: tested on macOS, Windows, Linux and Docker Containers
- Introduces an interactive shell-like CLI program to manage hunts and treasures via commands
- Uses logs to track user operations, with symlinked logs for centralized access
- Implements a file-based data storage system with binary records
- Utilizes multi-process architecture and sigaction-based signal handling for inter-process communication
- Enables runtime features such as live monitoring, hunt and treasure inspection, and controlled shutdown of the monitor process
- Designed and tested in a Unix (MacOS) environment using Clang, Make, and standard system calls
- Interactive Map: Mark and view safety alerts using Google Maps.
- Real-Time Notifications: Alerts for hazards in your vicinity.
- Automated Civic Complaints: Sends reports to local town halls.
- Community Trust System: Users vote to verify alerts.
- Frontend: Java (Android)
- Backend: Python (Django/Flask)
- Maps: Google Maps API
- Database: PostgreSQL
- Notifications: Firebase Cloud Messaging (FCM)
A video transmission and image processing system using FPGA, with real-time camera input and display via VGA. Supports basic image processing and integrates with OpenCV for face recognition through UART.
- FPGA: Xilinx Nexys 4 Artix-7
- Camera: OV7670
- IDE: AMD Vivado 2023
- Languages: Verilog / VHDL, Python
- Video output to monitor via VGA
- UART webcam feed for OpenCV-based face recognition
- Basic image processing on 12-bit RGB image
- Arduino UNO R4 WiFi
- I2C LCD Display
- DHT11 Temperature and Humidity Sensor
- Data Reading from Temperature and Humidity Sensors
- Data Retrieval using OpenWeather API Keys
- Data Output using an LCD Display or the on-board LED Dot Matrix