Form and Function
Design Form and Function Prediction From a Single Image
1MIT
Design Form and Function Prediction From a Single Image
1MIT
Estimating a design's form and functional performance in the early stages can be crucial for a designer for effective ideation. Humans have an innate ability to guess the size, shape, and type of a design from a single view. The brain fills in the unknowns in a fraction of a second. However, humans may struggle with estimating the performance of designs in the early stages of the design process without making prototypes or doing back-of-the-envelope calculations. In contrast, machines need information about the full 3D model of a design to understand its structure. Machines can estimate performance using pre-defined rules, expensive numerical simulations, or machine learning models. In this paper, we show how information about a design's form and functional performance can be estimated from a single image using machine learning methods. Specifically, we leverage the image-to-image translation method to predict multiple projections of an image-based design. We then train deep neural network models on the predicted projections to provide estimates of design performance. We demonstrate the effectiveness of our method by predicting the aerodynamic performance from images of aircraft models. To estimate ground truth aerodynamic performance, we run CFD simulations for 4045 3D aircraft models from the ShapeNet dataset and use their lift-to-drag ratio as the performance metric. Our results show that single images do carry information for both form and functional performance. We can produce six additional images of a design in different orientations from a single image, with an average Structural Similarity Index score of 0.872. We also find that image-translation methods provide a promising direction in estimating the performance of a design. Using multiple images of a design (gathered through image-translation) to predict design performance yields a recall value of 47%, 14% higher than a base guess and 3% higher than using a single image. Our work identifies the potential and provides a framework for using a single image to predict a design's form and functional performance during the early-stage design process.
Our architecture starts with a single input image of a design, from which we generate multiple images of this design at different orientations and finally estimate the aerodynamic performance of the design using machine learning.
Edwards, Kristen, Vaishnavi Addala, and Faez Ahmed. "Design Form and Function Prediction From a Single Image.” In Proceedings of the International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, {IDETC-21}, virtual, online, 2021.
@inproceeedings{edwards2021formfunction,
title={Design Form and Function Prediction From a Single Image},
author={Kristen M. Edwards and Vaishnavi L. Addala and Faez Ahmed},
year={2021},
booktitle={International Design Engineering Technical Conferences and Computers and Information in Engineering Conference,
{IDETC-21}},
organization={ASME},
day = {17-20},
month = {Aug},
address = {Virtual,
Online},
year={2021}}
We thank the Ida M. Green Fellowship for supporting Kristen Edwards’s research. We thank MIT’s Undergraduate Research Opportunities Program for supporting Vaishnavi Addala’s research. We also thank the other members of the DeCoDE Lab for helping at various stages of the project.
