Artificial Intelligence/Machine Learning, ASA(ALT), Direct to Phase II

Height of Burst Scoring through Machine Learning

Release Date: 11/16/2021
Solicitation: 21.4
Open Date: 11/30/2021
Topic Number: A214-047
Application Due Date: 01/04/2022
Duration: Up to 18 months
Close Date: 01/04/2022
Amount Up To: 1.7M

Objective

The purpose of this topic is to develop an AI/ML approach to score height of burst automatically, eliminating the man-power required by the current method. This is image processing to measure a distance on standard video files.

Description

The purpose of this topic is to create an automated, AI/ML-based technique to score height of burst testing for fuses and use AI/ML to estimate parameters in standard video files. Today height of burst tests are scored by human experts pausing video feeds of tests and estimating the height on the screen. Height of Burst (HOB) scoring is fully manual operation requiring pausing of video and measuring by an operator relative to the screen. AI/ML based Computer Vision (CV) Algorithms have been demonstrated to be robust for the purposes of estimating parameters in images and full motion video (FMV). AI/ML will dramatically accelerate this process by scoring in real-time. This new approach would train an AI to identify the height of burst based on previously scored data. If successful, the results from testing should be available in real time and cost significantly less in terms of man-hours.

Phase I

Direct to Phase 2 requires demonstration of AI/ML based computer vision algorithms for purposes of estimating parameters in images and full motion video, including pertinent data and report(s).

Phase II

Phase II will consist of identifying the HoB requirements for projectile use cases. Training algorithms on existing, verified data sets must also occur as well as demonstrating performance of algorithm on wide range of HoB experiments.

Phase III

Phase III should consist of deploying algorithm as an alternative in field test and quantify the reduction in time, cost of use, and performance. Transitioning to test ranges for use during proximity fuze and munition tests is also a required field for commercialization.

For more information, and to submit your full proposal package, visit the DSIP Portal.

References:

Measuring Distance Between Objects in an Image with OpenCV, Adrian Rosebrock, https://www.pyimagesearch.com/2016/04/04/measuring-distance-between-objects-in-an-image-with-opencv/

Vehicle Detection and Distance Estimation, Milutin N. Nikolic, https://towardsdatascience.com/vehicle-detection-and-distance-estimation-7acde48256e1

Learning Object-Specific Distance From a Monocular Image, Jing Zhu & Yi Fang, https://openaccess.thecvf.com/content_ICCV_2019/papers/Zhu_Learning_Object-Specific_Distance_From_a_Monocular_Image_ICCV_2019_paper.pdf

Objective

The purpose of this topic is to develop an AI/ML approach to score height of burst automatically, eliminating the man-power required by the current method. This is image processing to measure a distance on standard video files.

Description

The purpose of this topic is to create an automated, AI/ML-based technique to score height of burst testing for fuses and use AI/ML to estimate parameters in standard video files. Today height of burst tests are scored by human experts pausing video feeds of tests and estimating the height on the screen. Height of Burst (HOB) scoring is fully manual operation requiring pausing of video and measuring by an operator relative to the screen. AI/ML based Computer Vision (CV) Algorithms have been demonstrated to be robust for the purposes of estimating parameters in images and full motion video (FMV). AI/ML will dramatically accelerate this process by scoring in real-time. This new approach would train an AI to identify the height of burst based on previously scored data. If successful, the results from testing should be available in real time and cost significantly less in terms of man-hours.

Phase I

Direct to Phase 2 requires demonstration of AI/ML based computer vision algorithms for purposes of estimating parameters in images and full motion video, including pertinent data and report(s).

Phase II

Phase II will consist of identifying the HoB requirements for projectile use cases. Training algorithms on existing, verified data sets must also occur as well as demonstrating performance of algorithm on wide range of HoB experiments.

Phase III

Phase III should consist of deploying algorithm as an alternative in field test and quantify the reduction in time, cost of use, and performance. Transitioning to test ranges for use during proximity fuze and munition tests is also a required field for commercialization.

For more information, and to submit your full proposal package, visit the DSIP Portal.

References:

Measuring Distance Between Objects in an Image with OpenCV, Adrian Rosebrock, https://www.pyimagesearch.com/2016/04/04/measuring-distance-between-objects-in-an-image-with-opencv/

Vehicle Detection and Distance Estimation, Milutin N. Nikolic, https://towardsdatascience.com/vehicle-detection-and-distance-estimation-7acde48256e1

Learning Object-Specific Distance From a Monocular Image, Jing Zhu & Yi Fang, https://openaccess.thecvf.com/content_ICCV_2019/papers/Zhu_Learning_Object-Specific_Distance_From_a_Monocular_Image_ICCV_2019_paper.pdf

Height of Burst Scoring through Machine Learning

Scroll to Top