Detecting Potentially Life-Threatening Near Earth Objects in NASA Wise Data Using an Efficient Deep Neural Network
CSEF · 2023 Physics & Astronomy Honorable_mention Award
Overview
Over the last decade, astronomers have documented more than 90% of the Near Earth Objects (NEOs) larger than 1 km in diameter, but have missed detections of smaller NEOs. Small NEOs are harder to detect because of their size, but they are still capable of destroying entire cities. To solve this problem, I engineered a neural network designed to discover these faint NEOs. Images were obtained from NASA’s Wide-field Infrared Survey Explorer and then systematically injected with artificially generated NEOs. To improve detection capabilities of faint objects, this artificial population was skewed to have more dim NEOs (~6 SNR). The model was also trained on sets of images taken from several regions of space, allowing it to maintain high accuracy across the entire night sky. However, WISE’s images are full of NEO-like artifacts such as satellite trails or cosmic rays which succeeded in fooling last year’s initial prototype neural network. This year’s iteration includes a robust algorithm to preprocess all images, removing almost all such artifacts. The model was tested on a set of 2400 previously discovered NEOs and achieved 100% accuracy for faint NEOs, only missing 100 very bright objects. The model has already made several new detections NASA missed. The research here will supplement future asteroid mining and missions like DART which intends to destroy hazardous objects heading towards Earth. With the release of data from more powerful telescopes like the James Webb Space Telescope, the same methods here can be applied for even more faint detections.
Source coverage
This record comes from a published award list, not a complete project archive. Its abstract comes from CSEF's public project showcase as archived by the Internet Archive before judging (https://web.archive.org/web/20230401224130/https://ca-csef.zfairs.com/showcase/ShowcaseInfo?f=838e60b7-ea75-46e8-865c-fde4864244b3); the version presented may differ.
Awards (1)
- Category Award: HM
Competition history
- CSEF 2023
Resources
Related projects
ISEF · 2024
New Population of Asteroids Discovered! Rapidly Finding and Tracking Faint Near-Earth Asteroids Using an Accurate Deep Learning Based Process
ISEF · 2021
The Fast and Inconspicuous: New Near Earth Asteroids Discovered Using Deep Learning and Synthetic Data Are Fainter and Move Faster than Those Previously Discovered
CSEF · 2026
Autonomous Optical Satellite Location Using Low-Power Neural Architectures for Orbital Surveillance
CSEF · 2024
Dim Asteroids Discovered! Rapidly Finding and Tracking Faint Near-Earth Asteroids Using an Accurate Deep Learning Based Process
ISEF · 2025
Optimizing Exoplanet Detection: A Comparison of Neural Networks and Traditional Machine Learning Models on Light Curve Data
CWSF · 2026
Automating Asteroid Detection Criteria to Strengthen Citizen Science for Planetary Defense
ISEF · 2021
Detection and Characterization of Astronomical Dwarfs using CatWISE
JSHS · 2023
Using Deep Neural Networks to Identify Exoplanets
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: California Science & Engineering Fair public projects