Category recognition system using two ultrasonic sensors and combinational logic circuit

Moritake, Yusuke; Hikawa, Hiroomi

doi:10.1002/ecjc.20147

Cited by 5 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our target task of reverse vending machines, the object material is important. It has been shown that material classification is viable with ultrasonic signals [42,43], so we decided to use ultrasonic sensors as a new modality.…”

Section: Cmu-mosi and Cmu-moseimentioning

confidence: 99%

See 1 more Smart Citation

Correspondence Learning for Deep Multi-Modal Recognition and Fraud Detection

Park¹,

Kim

Choi

2021

Electronics

View full text Add to dashboard Cite

Deep learning-based methods have achieved good performance in various recognition benchmarks mostly by utilizing single modalities. As different modalities contain complementary information to each other, multi-modal based methods are proposed to implicitly utilize them. In this paper, we propose a simple technique, called correspondence learning (CL), which explicitly learns the relationship among multiple modalities. The multiple modalities in the data samples are randomly mixed among different samples. If the modalities are from the same sample (not mixed), then they have positive correspondence, and vice versa. CL is an auxiliary task for the model to predict the correspondence among modalities. The model is expected to extract information from each modality to check correspondence and achieve better representations in multi-modal recognition tasks. In this work, we first validate the proposed method in various multi-modal benchmarks including CMU Multimodal Opinion-Level Sentiment Intensity (CMU-MOSI) and CMU Multimodal Opinion Sentiment and Emotion Intensity (CMU-MOSEI) sentiment analysis datasets. In addition, we propose a fraud detection method using the learned correspondence among modalities. To validate this additional usage, we collect a multi-modal dataset for fraud detection using real-world samples for reverse vending machines.

show abstract

Section: Cmu-mosi and Cmu-moseimentioning

confidence: 99%

“…However, the experiments conducted in [42,43] were highly controlled in that the target objects were flat board shapes with the same pose and distance from the sensors. In real-world cases, the target objects would be in various shapes, sizes and poses.…”

Section: Cmu-mosi and Cmu-moseimentioning

confidence: 99%

Correspondence Learning for Deep Multi-Modal Recognition and Fraud Detection

Park¹,

Kim

Choi

2021

Electronics

View full text Add to dashboard Cite

show abstract

“…However, these methods are unable to detect dangerous ground surfaces. Furthermore, studies also examined material distinction characteristics [7,8]. The methods examined the reflected waveform as a 2D image, and introduced a pattern-matching approach based on benchmark patterns.…”

Section: Introductionmentioning

confidence: 99%

Application of Ultrasonic Sensors in Road Surface Condition Distinction Methods

Nakashima

Aramaki

Kitazono

et al. 2016

Sensors

View full text Add to dashboard Cite

The number of accidents involving elderly individuals has been increasing with the increase of the aging population, posing increasingly serious challenges. Most accidents are caused by reduced judgment and physical abilities, which lead to severe consequences. Therefore, studies on support systems for elderly and visually impaired people to improve the safety and quality of daily life are attracting considerable attention. In this study, a road surface condition distinction method using reflection intensities obtained by an ultrasonic sensor was proposed. The proposed method was applied to movement support systems for elderly and visually impaired individuals to detect dangerous road surfaces and give an alarm. The method did not perform well in previous studies of puddle detection, because the alert provided by the method did not enable users to avoid puddles. This study extended the method proposed by previous studies with respect to puddle detection ability. The findings indicate the effectiveness of the proposed method by considering four road surface conditions. The proposed method could detect puddle conditions. The effectiveness of the proposed method was verified in all four conditions, since users could differentiate between road surface conditions and classify the conditions as either safe or dangerous.

show abstract