Abstract
The demand for object detection capability in edge computing systems has surged. As such, the need for lightweight Convolutional Neural Network (CNN)-based object detection models has become a focal point. Current models are large in memory and deployment in edge devices is demanding. This shows that the models need to be optimized for the hardware without performance degradation. There exist several model compression methods; however, determining the most efficient method is of major concern. Our goal was to rank the performance of these methods using our application as a case study. We aimed to develop a real-time vehicle tracking system for cargo ships. To address this, we developed a weighted score-based ranking scheme that utilizes the model performance metrics. We demonstrated the effectiveness of this method by applying it on the baseline, compressed, and micro-CNN models trained on our dataset. The result showed that quantization is the most efficient compression method for the application, having the highest rank, with an average weighted score of 9.00, followed by binarization, having an average weighted score of 8.07. Our proposed method is extendable and can be used as a framework for the selection of suitable model compression methods for edge devices in different applications.
Highlights
We show the effectiveness of this scheme by applying it on the baseline, compressed, and micro-Convolutional Neural Network (CNN) models trained on our dataset
We evaluated and ranked the state-of-the-art methods for CNN model compression for resource-constrained edge devices using a weighted score-based ranking scheme that we developed
Our ranking method uses five key metrics computed for each model generated by the compression methods
Summary
Object classification tasks are solved using Convolutional Neural. CNNs are variants of Deep Neural Network (DNN) architectures that accept batches of images as input and return the probability vectors of all the possible outcomes [1]. These architectures are used as the backbone of state-of-the-art DNN-based object detection methods. R-CNN used AlexNet (a variant of the CNN architecture developed in [1], having over 62M trainable parameters and requiring a storage size of 250MB) as the backbone of the network. Other CNN architectures used as the backbone of object detection models are ResNet-50 [3], which requires over 95MB of storage space, and VGG16 [4]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
