Interior Designer Assistant
Publication Date: 2018-Aug-10
The IP.com Prior Art Database
Interior Designer Assistant
A person that is not trained in interior design might have difficulty determining whether an item (e.g., furniture, art work, flooring, etc.) matches the style of current furnishings, and consequently have difficulty selecting and purchasing new items. Many sources provide furnishing ideas based on style (e.g., home magazines, TV shows, websites, etc.) but referencing all these sources and figuring out how to apply the information is not a simple task.
The novel contribution is an interior design assistant that can predict whether a selected item will match a given set of items (e.g., a new piece of furniture for an existing/planned room), and then provide weighted percentages to represent the prediction. The novel system generates a database of reference items and associated tags (attributes) by category (room), and uses that database for attribute graph analysis.
The interior design assistant system identifies and ingests media sources of staged home styling recommendations. The system uses image recognition and Natural Language Processing (NLP) to identify the individual items and the associated characteristics within the media source. The user inputs a list of current furnishings, by room or set of rooms. When a user needs help deciding whether a new piece will stylistically fit within a specific room, the user first identifies the item in question. The system inputs/extracts that item’s characteristics. Based on the new item’s characteristics and the characteristics of the currently owned items, the system references an attribute graph to determine the style affinity strength of the new piece to the current furnishings.
The core components of the system include:
• User Device
− User interface that allows the user to interact with the design assistant
− User profile stores information about the furniture the user currently owns
− Camera for inputting images of new furniture to be purchased, existing rooms, etc.
• Media sources (e.g., websites, television shows, magazines, etc.)
− Content extraction module ingests media sources and outputs to a reference database
− Design assistant determines how well a new furniture item will fit within a user’s existing furnished room
Figure 1: Network diagram
Process Flow: Staged Furnishing Content Extraction
1. System ingests the media source into to the style characteristic extraction engine
2. Extraction engine (existing cognitive system application programming interfaces APIs)) generates tags from:
A. The tokens generated from visual recognition from the images and video
B. NLP of any text (from magazines) or audio to text (from video sources)
3. System classifies tags for each identified item
A. By item type (e....