This model is a fine-tuned version of the Qwen3 base model, specialized for road safety understanding and reasoning. It has been instruction-tuned on a custom dataset of 33,000 image-text conversations, where each example consists of an input image and a text prompt, and the model generates an appropriate context-aware response related to road safety.
