With the development of the field of e-nose research, the potential for application is increasing in various fields, such as leak measurement, environmental monitoring, and virtual reality. In this study, we characterize electronic nose data as structured data and investigate and analyze the learning efficiency and accuracy of deep learning models that use unstructured data. For this purpose, we use the MOX sensor dataset collected in a wind tunnel, which is one of the most popular public datasets in electronic nose research. Additionally, a gas detection platform was constructed using commercial sensors and embedded boards, and experimental data were collected in a hood environment such as used in chemical experiments. We investigated the accuracy and learning efficiency of deep learning models such as deep learning networks, convolutional neural networks, and long short-term memory, as well as boosting models, which are robust models on structured data, using both public and specially collected datasets. The results showed that the boosting models had a faster and more robust performance than deep learning series models.
1 February 2022 in Transaction of the Korean Society of Automotive Engineers
This paper presents a systematic comparative study of optical character recognition (OCR) architectures for Korean license plate recognition under identical detection conditions. Although recent automatic license plate recognition (ALPR) systems increasingly adopt Transformer-based decoders, it remains unclear whether performance differences arise primarily from sequence modeling strategies or from backbone feature representations. To address this issue, we employ a unified YOLOv12-based license plate detector and evaluate multiple OCR configurations, including a CNN with an Attention-LSTM decoder and a MobileNetV3 with a Transformer decoder. To ensure a fair comparison, a controlled ablation study is conducted in which the CNN backbone is fixed to ResNet-18 while varying only the sequence decoder. Experiments are performed on both static image datasets and tracking-based sequential datasets, assessing recognition accuracy, error characteristics, and processing speed across GPU and embedded platforms. The results demonstrate that the effectiveness of sequence decoders is highly dataset-dependent and strongly influenced by feature quality and region-of-interest (ROI) stability. Quantitative analysis further shows that tracking-induced error accumulation dominates OCR performance in sequential recognition scenarios. Moreover, Korean license plate–specific error patterns reveal failure modes not captured by generic OCR benchmarks. Finally, experiments on embedded platforms indicate that Transformer-based OCR models introduce significant computational and memory overhead, limiting their suitability for real-time deployment. These findings suggest that robust license plate recognition requires joint consideration of detection, tracking, and recognition rather than isolated optimization of OCR architectures.