Skip to main content
U.S. flag

An official website of the United States government

Dot gov

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Https

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Detection of Dense, Overlapping, Geometric Objects

Published

Author(s)

Adele P. Peskin, Boris Wilthan, Michael P. Majurski

Abstract

Using a unique data collection, we are able to study the detection of dense geometric objects in image data where object density, clarity, and size vary. The data is a large set of black and white images of scatterplots, taken from journals reporting thermophysical property data of metal systems, whose plot points are represented primarily by circles, triangles, and squares. We were able to build a highly accurate single class U-Net convolutional neural network model to capture 97% of image objects in a defined set of test images, locating the centers of the objects to within a few pixels of the correct locations. Developing a multiple class model that distinguishes the geometric shapes required a different ground truth annotation approach. As more geometric information is added to the segmentation masks, the outcomes of multi-class detections changed. We show that altering the ground truth annotations in the segmentation masks increases both the accuracy of object classification and localization on the plots, more than other factors such as adding loss terms to the network calculations.
Citation
International Journal of Artificial Intelligence and Applications
Volume
11

Keywords

object detection, convolutional neural networks, thermophysical properties

Citation

Peskin, A. , Wilthan, B. and Majurski, M. (2020), Detection of Dense, Overlapping, Geometric Objects, International Journal of Artificial Intelligence and Applications, [online], https://doi.org/10.5121/ijaia.2020.11403 29 (Accessed April 19, 2021)
Created June 30, 2020, Updated September 28, 2020