Adaptation of object detection networks under anomalous conditions

Out of distribution (OOD) samples can negatively affect the performance of deep neural networks. When deep neural networks are used in cyber-physical systems, it may be vulnerable to OOD data, leading to large errors and compromise the safety of the system. This paper proposes combining OOD explanat...

Full description

Saved in:
Bibliographic Details
Main Author: Koh, Rachel
Other Authors: Arvind Easwaran
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/166069
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-166069
record_format dspace
spelling sg-ntu-dr.10356-1660692023-04-21T15:38:52Z Adaptation of object detection networks under anomalous conditions Koh, Rachel Arvind Easwaran School of Computer Science and Engineering arvinde@ntu.edu.sg Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence Out of distribution (OOD) samples can negatively affect the performance of deep neural networks. When deep neural networks are used in cyber-physical systems, it may be vulnerable to OOD data, leading to large errors and compromise the safety of the system. This paper proposes combining OOD explanation and object detection model into a pipeline that maximizes efficiency and maintains accuracy across different OOD situations. We use YOLOv7 as the object detection model, which accepts different input sizes with a single set of weights. Under anomalous conditions, we can increase the input size to reduce the drop in accuracy at the expense of speed. This allows accuracy and speed to be balanced under different anomalous conditions. Alternatively, fine-tuned weights can be switched in under anomalous conditions, which shows consistent improvements though at higher costs. Bachelor of Engineering (Computer Science) 2023-04-19T01:15:53Z 2023-04-19T01:15:53Z 2023 Final Year Project (FYP) Koh, R. (2023). Adaptation of object detection networks under anomalous conditions. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/166069 https://hdl.handle.net/10356/166069 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
spellingShingle Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
Koh, Rachel
Adaptation of object detection networks under anomalous conditions
description Out of distribution (OOD) samples can negatively affect the performance of deep neural networks. When deep neural networks are used in cyber-physical systems, it may be vulnerable to OOD data, leading to large errors and compromise the safety of the system. This paper proposes combining OOD explanation and object detection model into a pipeline that maximizes efficiency and maintains accuracy across different OOD situations. We use YOLOv7 as the object detection model, which accepts different input sizes with a single set of weights. Under anomalous conditions, we can increase the input size to reduce the drop in accuracy at the expense of speed. This allows accuracy and speed to be balanced under different anomalous conditions. Alternatively, fine-tuned weights can be switched in under anomalous conditions, which shows consistent improvements though at higher costs.
author2 Arvind Easwaran
author_facet Arvind Easwaran
Koh, Rachel
format Final Year Project
author Koh, Rachel
author_sort Koh, Rachel
title Adaptation of object detection networks under anomalous conditions
title_short Adaptation of object detection networks under anomalous conditions
title_full Adaptation of object detection networks under anomalous conditions
title_fullStr Adaptation of object detection networks under anomalous conditions
title_full_unstemmed Adaptation of object detection networks under anomalous conditions
title_sort adaptation of object detection networks under anomalous conditions
publisher Nanyang Technological University
publishDate 2023
url https://hdl.handle.net/10356/166069
_version_ 1764208173931036672