Train YOLO + VLM with one command. Auto-generate vision-language training data from YOLO labels - no extra labeling needed.

A benchmark suite for lightweight generative multimodal Vision-Language Models, comparing ViLT and SmolVLM under resource-constrained inference environments. Demonstrates CPU-only deployment, model evaluation, and multimodal reasoning with images and text, highlighting practical GenAI engineering for real-world applications.

AI storyteller built on Qwen2-VL that transforms cultural imagery into authentic narratives using multimodal and progressive fine-tuning.

A Text-Conditioned Segmentation system using CLIPSeg and LoRA to automate industrial quality assurance. The model dynamically segments structural cracks and drywall joints based on natural language prompts (e.g., "segment wall crack").

A Survey on Trends, Challenges and Future Direction in Fire Management using Autonomous Unmanned Aerial Vehicles