MS-ALN: Multiscale Attention Learning Network for Pest Recognition

Feng, Fuxiang; Dong, Hanlin; Zhang, Youmei; Li, Bin

doi:10.1109/access.2022.3167397

Cited by 15 publications

(4 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although PCNet (Zheng et al, 2023) is a lightweight framework, it also adds a complex coordinate attention mechanism and feature fusion process, and its accuracy is not as good as our trained general CNN. Among CNN-based methods, only MS-ALN+DL (Feng et al, 2022) exceeded our general branch. Although it solves the problems of localization, occlusion, and class imbalance, the model is extremely complex and the obtained accuracy is still not enough for application.…”

Section: Comparison On Ip102mentioning

confidence: 98%

Pest recognition based on multi-image feature localization and adaptive filtering fusion

Chen,

Guo

et al. 2023

Front. Plant Sci.

View full text Add to dashboard Cite

Accurate recognition of pest categories is crucial for effective pest control. Due to issues such as the large variation in pest appearance, low data quality, and complex real-world environments, pest recognition poses challenges in practical applications. At present, many models have made great efforts on the real scene dataset IP102, but the highest recognition accuracy is only 75%. To improve pest recognition in practice, this paper proposes a multi-image fusion recognition method. Considering that farmers have easy access to data, the method performs fusion recognition on multiple images of the same pest instead of the conventional single image. Specifically, the method first uses convolutional neural network (CNN) to extract feature maps from these images. Then, an effective feature localization module (EFLM) captures the feature maps outputted by all blocks of the last convolutional stage of the CNN, marks the regions with large activation values as pest locations, and then integrates and crops them to obtain the localized features. Next, the adaptive filtering fusion module (AFFM) learns gate masks and selection masks for these features to eliminate interference from useless information, and uses the attention mechanism to select beneficial features for fusion. Finally, the classifier categorizes the fused features and the soft voting (SV) module integrates these results to obtain the final pest category. The principle of the model is activation value localization, feature filtering and fusion, and voting integration. The experimental results indicate that the proposed method can train high-performance feature extractors and classifiers, achieving recognition accuracy of 73.9%, 99.8%, and 99.7% on IP102, D0, and ETP, respectively, surpassing most single models. The results also show that thanks to the positive role of each module, the accuracy of multi-image fusion recognition reaches the state-of-the-art level of 96.1%, 100%, and 100% on IP102, D0, and ETP using 5, 2, and 2 images, respectively, which meets the requirements of practical applications. Additionally, we have developed a web application that applies our research findings in practice to assist farmers in reliable pest identification and drive the advancement of smart agriculture.

show abstract

Section: Comparison On Ip102mentioning

confidence: 98%

Pest recognition based on multi-image feature localization and adaptive filtering fusion

Chen,

Guo

et al. 2023

Front. Plant Sci.

View full text Add to dashboard Cite

show abstract

“…18 students (43 students in total) won awards in AI related competitions, and 5 students have published academic papers related to this course. Among them, there are 8 national competition awards, 10 provincial awards, and 2 Science Citation Index (SCI) papers [20,21]. This is an achievement that has not been achieved in previous years of teaching.…”

Section: Teaching Strategiesmentioning

confidence: 99%

Applied Outcome Based Education to Engineering Courses

2022

JERP

View full text Add to dashboard Cite

Outcome-based education (OBE) is education in which an emphasis is placed on a clearly articulated idea of what students are expected to know and be able to do, that is, what skills and knowledge they need to have, when they leave the school system. This paper takes "Neural network and Deep Learning" as an example to present the practical process and educational outcomes of OBE on university's engineering courses. Separately, the practical process takes the curriculum context as the main line and educational outcomes are based on competition awards and scientific research achievements.

show abstract

“…Despite recent progress in the CNN and ViT families, pest image identification still suffers from problems such as complex backgrounds [22,23,24,25,26] and small sizes [2,17,27]. For accurate pest image identification, it is crucial to separate regions of interest (ROIs), such as wings and heads, from the background, which contains clues for pest identification.…”

Section: Main Issue For Pest Image Identificationmentioning

confidence: 99%

Making ‘the National Image’ of Korea: From the Shrine of the Joseon Dynasty to the National Memorial of the Republic of Korea

Kim

2022

Buildings

View full text Add to dashboard Cite

Korea tried to reconstruct its tradition as a modern nation and promote nationalism only in the 1960s, a relatively late period compared to the West. So-called “tradition-making” began by promoting the full-fledged national shrine project to commemorate historical figures in a short period. These national shrines were built in a traditional style and embodied a national image representing Korea. This study analyzed which elements of traditional Korean architecture were selectively adopted for the main hall of Hyeonchungsa, the first national shrine, and discussed how these choices were intertwined with the discourses on Korean architecture at the time by tracing their historical, cultural, and political context. Although the ‘newly invented’ national shrine functionally resembled a Confucian shrine, it favored splendid and magnificent elements on its exterior, a tendency that is well-demonstrated in the shape of its roof, the decoration of the eaves, and the elevation. This style reflects the modern perspective that regarded the roof curve as an essential feature to define the identity of Korean architecture. Additionally, it can be seen as an effort to reevaluate the architectural style of the Joseon Dynasty, which was belittled during the Japanese colonial period. In addition, these national shrines showed an attempt to reproduce the architectural form of the past with modern material–concrete–by actively referring to the drawing data derived from the actual measurement surveys of historical buildings that were carried out vigorously in the early 1960s. Although these buildings have not been valued in academia amid criticism of the political use of traditions, they played a critical role in spreading the “image of Korean tradition”.

show abstract

MS-ALN: Multiscale Attention Learning Network for Pest Recognition

Cited by 15 publications

References 34 publications

Pest recognition based on multi-image feature localization and adaptive filtering fusion

Pest recognition based on multi-image feature localization and adaptive filtering fusion

Applied Outcome Based Education to Engineering Courses

Making ‘the National Image’ of Korea: From the Shrine of the Joseon Dynasty to the National Memorial of the Republic of Korea

Contact Info

Product

Resources

About