ARCHIVES

Interior renovation planning usually needs repeated designer consultations, manual measurements and informal cost estimates. Many AI design tools generate attractive images from a single photo but ignore the real room geometry and give no budget. This paper proposes DecorAI, an end-to-end system that converts a short room video into a style-specific, photorealistic interior design and an itemized cost estimate. In the proposed design, key frames are extracted using OpenCV, and room geometry is recovered using monocular depth estimation and Structure from Motion (SfM). Depth and edge maps condition ControlNet, which guides Stable Diffusion XL (SDXL) with style-specific Low-Rank Adaptation (LoRA) weights so that walls, doors, windows and perspective are preserved. A fine-tuned YOLOv8 model detects furniture and décor, discarding low-confidence detections, and the detected objects are matched to a MySQL price database to produce minimum and maximum budget ranges. The system is organized into four modules: a dashboard, video and structure analysis, AI design and object analysis, and cost estimation and reporting. It is planned for implementation in Python with PyTorch, Hugging Face Diffusers and FastAPI. This paper presents the architecture, methodology and planned evaluation of the system.