This project is an Edge AI system that performs real-time object detection, audio transcription, and summarization of transcribed notes, all on-device using ONNX models. The goal is to enable low-latency AI tasks on mobile or edge hardware.
Real-time object detection using YOLOv8 models exported to ONNX.
Real-time audio recording and transcription using a Whisper model (Faster-Whisper) in ONNX.
Summarization of transcribed text using a BART model exported to ONNX.
CPU-only inference for broad device compatibility.
Python
ONNX Runtime
YOLOv8 / YOLOv8n models
Faster-Whisper
BART Summarization