VisioNiX bridges traditional computer vision with LLM-based reasoning: it detects and captions objects in an image (YOLOv8), extracts printed and handwritten text (OCR, multilingual), and combines all of it with a vision-language model (LLaVA / BLIP-2) to answer questions about what's in the image and hold a conversation grounded in it.
Python · YOLOv8 · OCR · LLMs · Team project
VisioNiX
A multimodal AI assistant that combines computer vision, OCR, and large language models to understand images and hold context-aware conversations about them. Built with a 4-person team.
Built collaboratively with Aryan Kate, Bhuvan Patil, and Siddhesh Patil.
Overview
Capabilities
- Object detection and scene analysis via YOLOv8
- Multilingual OCR for printed and handwritten text
- Multimodal reasoning combining image, text, and detected objects
- Chat-based interface for visual question answering
- UI-based fine-tuning for domain-specific use cases, with auto-deployment to Hugging Face