Litcius/Paper detail

Choice of Activation Function in Convolutional Neural Networks for Person Re-Identification in Video Surveillance Systems

Huafeng Chen, S. A. Ihnatsyeva, Rykhard Bohush, Sergey Ablameyko

2022Programming and Computer Software13 citationsDOI

Abstract

In this paper, we improve the accuracy of person re-identification in images obtained from distributed video surveillance systems by choosing activation functions for convolutional neural networks. The most popular activation functions used for object detection, namely, ReLU, Leaky-ReLU, PReLU, RReLU, ELU, SELU, GELU, Swish, and Mish, are analyzed based on the following metrics: Rank1, Rank5, Rank10, mAP, and training time. For feature extraction, ResNet-50, DenseNet-121, and DarkNet-53 architectures are employed. The experimental study is carried out on open datasets Market1501 and PolReID. The accuracy of person re-identification is assessed after thrice-repeated training and testing with different activation functions, neural network architectures, and datasets by averaging the values of the selected metrics.

Topics & Concepts

Convolutional neural networkComputer scienceActivation functionIdentification (biology)Artificial intelligencePattern recognition (psychology)Artificial neural networkFeature extractionObject (grammar)Object detectionMachine learningBotanyBiologyVideo Surveillance and Tracking MethodsAdvanced Neural Network ApplicationsFire Detection and Safety Systems
Choice of Activation Function in Convolutional Neural Networks for Person Re-Identification in Video Surveillance Systems | Litcius