Advanced Designing Assistant System for Smart Design Based on Product Image Dataset

Existing product images are very important references for designing a new scheme. However, the designers have to collect and organize the product image data manually without proper tools, which may be time-consuming, inefficient and expensive. The rapid growth of product design has called for a smart system to assist designers with a quick start in designing a new product. Therefore, we propose an advanced designing assistant system (ADAS) to help the designers handle the large-volume product images more efficiently and create better design. The ADAS utilizes big data and artificial intelligence technology to achieve mass product data acquisition, analysis, retrieval, and design scheme generation. The ADAS utilizes builds a product image dataset firstly to decrease high cost of time and money in images collection task. Furthermore, based on this dataset, the ADAS develops three applications: (1) image retrieval and infringement analysis, (2) multi-label semantic annotation, (3) automatic design scheme generation. Experiments are conducted to validate the merits of the proposed system. And the results show that the ADAS could support designers with high quality from initial data collection to image retrieval, infringement analysis, semantic learning, and design scheme generation throughout the entire flow of the design task, greatly shortening the design period and improving efficiency.