Accelerating Strawberry Ripeness Classification Using a Convolution-Based Feature Extractor along with an Edge AI Processor

Joungmin Park,Jinyoung Shin,Raehyeong Kim,Seongmo An,Sangho Lee,Jinyeol Kim,Jongwon Oh,Youngwoo Jeong,Soohee Kim,Yue Ri Jeong,Seung Eun Lee

DOI: https://doi.org/10.3390/electronics13020344

IF: 2.9

2024-01-14

Electronics

Abstract:Image analysis-based artificial intelligence (AI) models leveraging convolutional neural networks (CNN) take a significant role in evaluating the ripeness of strawberry, contributing to the maximization of productivity. However, the convolution, which constitutes the majority of the CNN models, imposes significant computational burdens. Additionally, the dense operations in the fully connected (FC) layer necessitate a vast number of parameters and entail extensive external memory access. Therefore, reducing the computational burden of convolution operations and alleviating memory overhead is essential in embedded environment. In this paper, we propose a strawberry ripeness classification system utilizing a convolution-based feature extractor (CoFEx) for accelerating convolution operations and an edge AI processor, Intellino, for replacing FC layer operations. We accelerated feature map extraction utilizing the CoFEx constructed with systolic array (SA) and alleviated the computational burden and memory overhead associated with the FC layer operations by replacing them with the k-nearest neighbors (k-NN) algorithm. The CoFEx and the Intellino both were designed with Verilog HDL and implemented on a field-programmable gate array (FPGA). The proposed system achieved a high precision of 93.4%, recall of 93.3%, and F1 score of 0.933. Therefore, we demonstrated a feasibility of the strawberry ripeness classification system operating in an embedded environment.

engineering, electrical & electronic,computer science, information systems,physics, applied

What problem does this paper attempt to address?

The paper aims to address two main issues in strawberry maturity classification:

1. **Reducing the computational burden of convolution operations**: Convolutional Neural Networks (CNNs) play a crucial role in assessing strawberry maturity, but convolution operations constitute the primary computational burden of CNN models.

2. **Alleviating the memory overhead of fully connected layers**: Fully connected (FC) layers require a large number of parameters and frequent external memory access, which is a challenge in embedded environments.

To address these issues, the paper proposes a strawberry maturity classification system based on a Convolutional Feature Extractor (CoFEx) and an Edge AI Processor (Intellino). Specifically:

- **Convolutional Feature Extractor (CoFEx)**: Utilizes an array-based architecture (SA) to accelerate convolution operations, efficiently extracting feature maps from images.

- **Edge AI Processor (Intellino)**: Employs the k-Nearest Neighbors (k-NN) algorithm instead of fully connected layer operations to reduce computational load and lower memory overhead.

Through these techniques, the paper demonstrates that the proposed system can achieve efficient strawberry maturity classification in embedded environments, achieving 93.4% accuracy, 93.3% recall, and an F1 score of 0.933.