Python on Zynq FPGA for Convolutional Neural Networks

Binary classification of images with a CNN developed for the Zynq-7000 SoC XC7Z020-1CLG400C FPGA available on the PYNQ-Z2 board

Implemented a basic convolutional neural network for performing real-time doodle classification on a Xilinx PYNQ P1 FPGA Board.

No description available

This project implements a Graph Fourier Transform (GFT)-based feature extraction pipeline for EEG signal classification using the ASU imagined speech dataset. The complete processing pipeline has been deployed and tested on a PYNQ-ZU FPGA board, achieving a classification accuracy of 87.23% using an SVM (Support Vector Machine) classifier.

Real-time bitstream malware detection using an ML model. Includes feature extraction, Trojan detection, family classification, and embedded inference on PYNQ-supported FPGA boards.

Hardware Accelerated QNN for Arrhythmia classification on Edge devices

Low-Precision Neural Networks for Classification on PYNQ with FINN

This project accelerates the convolutional layer of the AlexNet model on FPGA using the Vitis-Vivado-PYNQ framework, optimizing performance for efficient image classification and exploring future possibilities for FPGA-based AI applications.

FPGA-based CNN accelerator on PYNQ-Z2 using custom Verilog CNN layer modules integrated via AXI interfaces. Accelerates real-time image classification with Python-controlled overlay. Efficient hardware-software co-design reduces latency and boosts edge AI inference performance.

No description available

No description available

Code used for my Master's project.

First project on the PYNQ Z2 FPGA. First time experimenting with HW optimization - Late Github Upload

Supervised emotion classification using MuseBox and PYNQ

This repository provides the complete implementation of the SVM-based crop yield classification system, including model training, evaluation, and deployment on the PYNQ-Z2 platform. The code is shared to support reproducibility of the results presented in the paper.

Real Time Object Detection

Real-time object detection and classification using a CNN accelerator deployed on a PYNQ-Z2 FPGA, trained on the CIFAR-10 dataset and implemented using Vitis HLS and Vivado.

Real-time bitstream malware detection using an ML + DL model trained on SOTA benchmarks, sourced via Trust-Hub. Includes feature extraction, Trojan detection, family classification, and embedded inference on PYNQ-supported FPGA boards.

Real-time Edge AI on Kria KR260 using a custom DPU (B4096). Implements ResNet-50 classification with PetaLinux2022.2 + Vitis AI 3.0 and YOLOX-Nano object detection using DPU-PYNQ +Vitis AI 3.5 on Ubuntu2022. Demonstrates efficient, low-latency deep learning inference for robotic vision.

This project implements real-time arrhythmia detection on the edge using a 1D-CNN model. We first trained and validated the model through ML simulation for accurate ECG arrhythmia classification. The optimized model was then deployed on the PYNQ-Z2 FPGA board, enabling low-latency, low-power inference .