Multimodal Odour and Image Recognition on Synthetic Dataset

Fanny Monori; Alin-Sasa Tisan

Multimodal Odour and Image Recognition on Synthetic Dataset

Research output: Contribution to conference › Paper › peer-review

Abstract

Most odour recognition algorithms rely only on data from chemical (gas) sensors. There are scenarios in which it would help to include other modalities (data from other types of sensors) in the recognition. In this paper we describe a process of combining gas sensor data and image data for binary classification. For that, a synthetic dataset was assembled, and different neural networks were trained. We found that combining the modalities makes the classification perform better and be
more stable. The next step is the collection of multimodal data and validating this multimodal approach on a non-synthetic dataset

Original language	English
Publication status	Published - Jun 2023
Event	2023 IEEE 32nd International Symposium on Industrial Electronics (ISIE) - Duration: 17 Jul 2023 → …

Conference

Conference	2023 IEEE 32nd International Symposium on Industrial Electronics (ISIE)
Period	17/07/23 → …

Cite this

@conference{f2085fd934f64d60abe2ca009c2250c5,

title = "Multimodal Odour and Image Recognition on Synthetic Dataset",

abstract = "Most odour recognition algorithms rely only on data from chemical (gas) sensors. There are scenarios in which it would help to include other modalities (data from other types of sensors) in the recognition. In this paper we describe a process of combining gas sensor data and image data for binary classification. For that, a synthetic dataset was assembled, and different neural networks were trained. We found that combining the modalities makes the classification perform better and bemore stable. The next step is the collection of multimodal data and validating this multimodal approach on a non-synthetic dataset",

author = "Fanny Monori and Alin-Sasa Tisan",

year = "2023",

month = jun,

language = "English",

note = "<br/>2023 IEEE 32nd International Symposium on Industrial Electronics (ISIE) ; Conference date: 17-07-2023",

}

TY - CONF

T1 - Multimodal Odour and Image Recognition on Synthetic Dataset

AU - Monori, Fanny

AU - Tisan, Alin-Sasa

PY - 2023/6

Y1 - 2023/6

N2 - Most odour recognition algorithms rely only on data from chemical (gas) sensors. There are scenarios in which it would help to include other modalities (data from other types of sensors) in the recognition. In this paper we describe a process of combining gas sensor data and image data for binary classification. For that, a synthetic dataset was assembled, and different neural networks were trained. We found that combining the modalities makes the classification perform better and bemore stable. The next step is the collection of multimodal data and validating this multimodal approach on a non-synthetic dataset

AB - Most odour recognition algorithms rely only on data from chemical (gas) sensors. There are scenarios in which it would help to include other modalities (data from other types of sensors) in the recognition. In this paper we describe a process of combining gas sensor data and image data for binary classification. For that, a synthetic dataset was assembled, and different neural networks were trained. We found that combining the modalities makes the classification perform better and bemore stable. The next step is the collection of multimodal data and validating this multimodal approach on a non-synthetic dataset

M3 - Paper

T2 - <br/>2023 IEEE 32nd International Symposium on Industrial Electronics (ISIE)

Y2 - 17 July 2023

ER -