A Proper Orthogonal Decomposition approach for parameters reduction of Single Shot Detector networks

TL;DR

This paper introduces a Proper Orthogonal Decomposition-based framework to reduce hyperparameters in SSD networks, leading to smaller models and faster fine-tuning, especially beneficial for resource-limited embedded systems.

Contribution

It applies Proper Orthogonal Decomposition to compress SSD networks, reducing hyperparameters and improving transfer learning efficiency.

Findings

Significant reduction in network dimension.

Remarkable speedup in fine-tuning process.

Effective application to SSD300 on PASCAL VOC.

Abstract

As a major breakthrough in artificial intelligence and deep learning, Convolutional Neural Networks have achieved an impressive success in solving many problems in several fields including computer vision and image processing. Real-time performance, robustness of algorithms and fast training processes remain open problems in these contexts. In addition object recognition and detection are challenging tasks for resource-constrained embedded systems, commonly used in the industrial sector. To overcome these issues, we propose a dimensionality reduction framework based on Proper Orthogonal Decomposition, a classical model order reduction technique, in order to gain a reduction in the number of hyperparameters of the net. We have applied such framework to SSD300 architecture using PASCAL VOC dataset, demonstrating a reduction of the network dimension and a remarkable speedup in the fine-tuning of the network in a transfer learning context.