Insect Pest Image Recognition: A Few-Shot Machine Learning Approach including Maturity Stages Classification

Abstract

Recognizing insect pests using images is an important and challenging research issue. A correct species classification will help choosing a more proper mitigation strategy regarding crop management, but designing an automated solution is also difficult due to the high similarity between species at similar maturity stages. This research proposes a solution to this problem using a few-shot learning approach. First, a novel insect data set based on curated images from IP102 is presented. The IP-FSL data set is composed of 97 classes of adult insect images, and 45 classes of early stages, totalling 6817 images. Second, a few-shot prototypical network is proposed based on a comparison with other state-of-art models and further divergence analysis. Experiments were conducted separating the adult classes and the early stages into different groups. The best results achieved an accuracy of 86.33% for the adults, and 87.91% for early stages, both using a Kullback–Leibler divergence measure. These results are promising regarding a crop scenario where the more significant pests are few and it is important to detect them at earlier stages. Further research directions would be in evaluating a similar approach in particular crop ecosystems, and testing cross-domains.

BAU-Insectv2: An agricultural plant insect dataset for deep learning and biomedical image analysis

“…This dataset facilitates the creation and assessment of machine learning and deep learning models [2] aimed at insect classification [3] , pest detection [4] , and plant protection. The diverse sample sizes per genus offer a realistic and challenging environment for both training and testing models, mirroring the real-world distribution of insect species in agricultural contexts.…”

Non-Destructive Detection of Soybean Pest Based on Hyperspectral Image and Attention-ResNet Meta-Learning Model

“…Meta-learning algorithms can be divided into three categories: model-based algorithms, metric-based algorithms, and optimization-based algorithms. Gomes J C et al [ 14 ] explored a prototype model of a few-shot, and used Kullback-Leibler divergence measurement to detect pests in the mature stage and early stage. Finally, the accuracy rate in the mature stage was 86.33%, and that in the early stage was 87.91%.…”

Insect recognition based on complementary features from multiple views

“…Many advanced studies have been implemented on this dataset, which obtained improved performances in varying degrees 6 , 10 , 11 or less computation time 12 . It is worth mentioning that Gomes et al curated the images from IP102 to compromise two datasets composed of adult insect and early-stage insect images 13 . They got significantly higher accuracy than the work implemented on the original IP102 dataset that mixed multiple growth stages in a specific class, demonstrating that different growth forms indeed increase the difficulty of recognition.…”