Recognition of Rhyzopertha dominica adults by digital imaging was first published by Zayas and Flinn (1998) and a second paper recognizing Oryazephilus surinamensis adults by digital imaging was soon published by Ridgway et al. (2002). Tadiparthi et al. (2027) reviewed 29 studies using supervised learning to detect or identify grain infesting insects to species. These included three kinds of studies: elytra only (Bisgin et al. 2018, 2022, Park et al. 2016, Wu et al. 2019), x-ray images of internal feeding insects (Barboza da Silva et al. 2021, Boniecki et al. 2014, Divyanth et al. 2022, ) and live adult insects often on complex grain background (Agarwal et al. 2020, Badgujar et al 2023a, b, Chen et al. 2022, Chu et al 2023, Li et al. 2019, Li et al. 2020, 2022, Li et al. 2024b, Liu et al. 2018, Lyu et al. 2021a, Mendoza et al. 2023, Qi et al. 2025, Shen et al. 2018, Shi et al. 2020, Tian et al. 2025, Yang et al. 2022, Zhang et al. 2021, Zhao et al. 2024a, Zhou et al. 2019, Zhu et al. 2025). YOLO and Faster R-CNN variants were the most widely adopted and modified architectures, while recent studies increasingly incorporated attention and transformer-based components, backbone modifications, multi-scale feature extraction, feature pyramid and feature fusion networks to improve small-insect detection. Small insect size, occlusion, overlapping instances, motion blur, illumination variability, dust, cluttered backgrounds, and similarity between insects and grain surfaces continue to affect recognition accuracy. Martin et al. (2016), Bisgin et al. (2018) and Wu et al. (2019) reused elytra images from Park et al. (2016), and Tian et al. (2025) and Zhu et al. (2025) reused images from Li et al. 2024b. Only 4 of 29 studies released public datasets (*Bisgin et al., (2022), Li et al. (2019, 2024) Park et al., 2016) and only three demonstrated edge (**Chen et al. 2022, Mendoza et al. 2023) or mobile deployment (**Badgujar et al. 2023b). Chen et al. 2022 used a mobile robot to locate insects. Two of these 29 studies counted insects (Qi et al. 2025, Zhu et al. 2025). Li et al. (2024a) was able use deep learning to accurately count Lasioderma serricorne. Machine learning identified species and geographical strains of Sitophilus oryzae and Sitophilus zeamais (Cao et al. 2015) and sounds of five insect species moving and feeding have been used to identify them to species (Balingbing et al. 2024, Banga et al. 2020, Montemayor et al 2024).
Sitophilus oryzae was included in 20 of 29 studies (Tadiparthi et al. 2027), Oryzaephilus surinamensis and Rhyzopertha dominica each in 19, Tribolium castaneum in 18, Cryptolestes ferrugineus in 14, Sitophilus zeamais in 13, Lasioderma serricorne in 12, Cryptolestes pusillus in 9, Tribolium confusum in 8, Stegobium paniceum and Sitophilus granarius each in 5, Cryptolestes turcicus, Gnathocerus cornutus, Oryzaephilus mercator and Tribolium madens each in 4, Plodia interpunctella and Trogoderma variabile each in 3 and Ahasverus advena, Callosobruchus maculatus, Cynaeus angustus, Sitotroga cerealella, Tribolium brevicornis, Tribolium freeman and Zabrotes subfasciatus each in 2. Another 11 species were included in only one study (Araecerus fasciculatus (<80%), Attagenus unicolor (<80%), Cathartus quadricollis (<80%), Ephestia elutella (92.4%), Gibbium aequinoctiale (97.3%), Latheticus oryzae (>80%), Lophocateres pusillus (>80%), Palorus ratzeburgii (>80%), Tribolium destructor >80%), Trogoderma granarium (>90%) and Trogoderma inclusum (>80%). Tuda and Luna-Maldonado (2020) found that image analysis can determine both species and sex of Callosobruchus chinensis and two of its parasitoids, Anisopteromalus calandrae and Heterospilus prosopidis. Data have been collected for 38 stored-product insect species.
Li et al. (2026a) summarized progress and challenges in research on detection of stored-grain insect pests based on deep learning covering many additional studies including six dissertations (Guo 2023, Liang 2024, Lu 2024, Wang 2024a, Wang 2024b, Zhang 2024a). FCA-YOLO, based on YOLOv8, achieves 4.51% improvement in accuracy of pest detection and inspection trolleys were deployed to achieve real-time prediction and analysis of pests in the warehouse (Ge 2025). Replacing Darknet—53 with Shufflenetv2, a lighter network, in YOLOv8m algorithm, adding Squeeze—and—Excitation Networks attention mechanism at the end of the backbone network to obtain high-quality feature maps, and adopting WIoUv3 Loss function improved accuracy and speed of insect detection (Lü et al. 2025). PDA-YOLO, an algorithm based on YOLO11n, integrates three key modules to improve stored-grain insect pest detection: PoolFormer_C3k2 (PF_C3k2) for efficient local feature extraction, Attention-based Intra-Scale Feature Interaction (AIFI) for enhanced global context awareness, and Dynamic Multi-scale Aware Edge (DMAE) for precise boundary detection, achieving mAP@0.5 of 96.6% (Sun et al. 2025). Faster R-CNN algorithm was improved by pyramid pooling module handling different scales of insects with white paper or grain background. (Xu et al. 2022). Spatial and channel attention mechanisms added to CSP (Cross Stage Partial) modules of YOLOv5s to facilitate learning of informative features and repeated residual units in the CSP module are removed to make more lightweight (Du et al. 2023). The higher accuracy for Sitotroga cerealella using Single Shot MultiBox Detector (SSD) can be attributed to its larger size than other pest species (Deng et al. 2020). Gaussian Mixture Model (GMM) is employed to capture moving pest images from video and three-frame differencing method generates a complete binary image of the pests which is identified by ResNet18 architecture (Zhang et al. 2024b). Model predicts need and estimates efficacy for insect control by fumigation or controlled-atmosphere. Grain loss caused by pests during storage at the Jinan No. 3 Grain Depot was reduced by 56% compared to previous levels, and storage costs were lowered by 35%.
The following studies are more recent. The enhanced support vector machine classification (SVC) model achieved 100 % accuracy for Sitophilus oryzae while reducing the predictive time (Yu et al. 2025). Gradient-weighted class activation mapping (Grad-CAM) indicates attention-assisted VGG7 model can mine important information in the image more accurately and quickly. Achieving accurate detection through object detection methods remains challenging due to small target scales, subtle inter-species morphological differences, and complex background interference and improvements with MSGP-DETR achieves 94.6 % AP50 accuracy (Guan et al. 2026). ABF-YOLO11 excels at distinguishing similar pest species and performed well with complex backgrounds (wheat, corn, paddy), overall accuracy of 94.4% accuracy (Hou et al. 2026). Swin Transformer and BiFPN are added to YOLOv8n to obtain better perception ability, positioning accuracy, detail retention and edge clarity, improving the recognition accuracy to 97.7% (Li et al. 2026b). Pest-YOLO increased precision of insect detection by 3.9 % compared with 10 published models and it was successfully deployed on edge computing platform (Sun et al. 2026). Deployment evaluations of YOLO-DInsects model based on the YOLOv11n on edge device confirmed its capability for real-time insect detection under resource-constrained field environments (Li et al 2026b). Yu et al. (2026) evaluates body surfaces texture features to enhance classification accuracy. OS-YOLO achieved a precision of 96.1% outperforming several state-of-the-art detection models for five major insect pests of Chinese medical material (Zhao et al. 2026).
In addition to monitoring gas (O2, CO2, PH3) (Jiang et al 2026) or temperature (Han and Leng 2026, Zhang et al. 2026, Zhang et al. 2027), artificial intelligence and machine learning have been used to forecast insect problems (Balingbing et al. 2024, Banga et al. 2020, Chen et al. 2026, Nyabako et al. 2020, Ogungbite and Ogungbite 2026, Siaho et al. 2023, Wu et al. 2016) and improve pest management with aeration (Junior et al, 2024, Luo et al. 2026, Zhang et al. 2026), heat disinfestation (Abdelsamea et al. 2023, Adler et al. 2026a, b, Rossos et al. 2025) and botanical insecticide (Casadei et al. 2026, Erdogan et al. 2026). Majewski (2022) used machine learning to develop a monitoring system for rearing Tenebrio molitor as an edible insects. Machine learning has been used to predict phototaxis of almond moth, Cadra cautella (Guru et al. 2025) and the life history of Callosobruchus chinensis (Gu and Tuda 2026).
Artificial neural networks were able to optimize moisture content during sesame seed drying (Haddad et al. 2025). Coupled with the Internet of Things remote monitoring is possible, because equipment for this is commercially available (Aivision 2025, Anonymous 2018, Direct-to-satellite IoT sensors 2026, HARDWARIO and GrainLink 2026, TeleSense 2026). Other reviews include Abdullah (2026), Adeyeye (2027), Anukiruthika and Jayas. (2025) and Devi and Singhrova (2026).
References (For the 29 studies reviewed by Tadiparthi et al. (2027), the numbers of species studied by each are given in front of author names. When more than one training method is used, the highest average accuracy or precision is given for each species in annotation).
Abdelsamea, Mohammed M., Mohamed Medhat Gaber, Aliyuda Ali, Marios Kyriakou, and Shams Fawki 2023. A Logarithmically Amortising Temperature Effect for Supervised Learning of Wheat Solar Disinfestation of Rice Weevil Sitophilus oryzae (Coleoptera: Curculionidae) Using Plastic Bags. Scientific Reports 13(1), 2655. Predicts whether a certain combination of parameters will be effective in the treatment of insects using thermal control. Random forest model achieved 99.5% accuracy in predicting lethal temperatures for pest control.
Abdullah, Hafiz Muhammad, Farhan Saeed, Muhammad Bilal Hussain, Ali Raza, Amar Shankar, Atreyi Pramanik, Gunjan Garg et al. 2026. Use of Artificial Intelligence in post-harvest losses management: a current insight. European Food Research and Technology 252(8), 305.
Adeyeye, Samuel Ayofemi Olalekan 2027. From reactive to predictive stored-product protection: AI, digital twins, biosensors, and multi-omics for smart grain storage J. Stored Prod. Res. 120, 103220.
Adler, Cornel, Gunnar Böttger, Kirko Große, Christian Hentschel, Dirk Höpfner, and Peter Kern 2026a. Artificial intelligence for stored product insect detection and control with laser beams: the Insect Laser project. Journal of Plant Diseases and Protection 133(3): 85.
Adler, Cornel, Gunnar Böttger, Kirko Große, Christian Hentschel, Dirk Höpfner, and Peter Kern 2026b. Artificial intelligence for stored product insect detection and control with laser beams: the Insect Laser project. Journal of Plant Diseases and Protection 133(3), 85. https://doi.org/10.1007/s41348-026-01273-8. Deep learning trained AI to recognize Plodia interpunctella and Sitophilus granarius with 93% confidence. Using laser on track to kill identified insects on grain surface before they reproduce could slow population growth and dead insects can be removed by normal grain cleaning.
2 Agarwal, M., Al-Shuwaili, T., Nugaliyadde, A., Wang, P., Wong, K.W., Ren, Y. 2020. Identification and diagnosis of whole body and fragments of Trogoderma granarium and Trogoderma variabile using visible near infrared hyperspectral imaging technique coupled with deep learning. Comput. Electron. Agric. 173, 105438. https://doi.org/ 10.1016/j.compag.2020.105438 Both >90%. (Bold only in this study).
Aivision 2025. Wireless smartprobe systems for smart, data-driven IPM practices. www. aivisionfood.comon
Anonymous 2018. Internet of Things Revolutionizes Farming and Grain Storage. Feb 20, 2018 Advanced grain monitoring systems like those available from Tri-States Grain Conditioning rely on the IoT. https://tsgcinc.com/internet-of-things-grain-temperature-monitoring/
Anukiruthika, T., and D. S. Jayas 2025. AI-Driven Grain Storage Solutions: Exploring Current Technologies, Applications, and Future Trends. J. Stored Prod. Res. 111, 102588.
5 Badgujar, C.M., Armstrong, P.R., Gerken, A.R., Pordesimo, L.O. and Campbell, J.F. 2023a. Identifying common stored product insects using automated deep learning methods. J. Stored Prod. Res. 103, 102166. https://doi.org/10.1016/j.jspr.2023.102166. Cryptolestes ferrugineus 100% = Rhyzopertha dominica 100% = Tribolium castaneum 100% > Sitophilus oryzae 99.9% > Oryzaephilus surinamensis 99.3%. Gradient-weighted Class Activation Mapping (Grad-CAM) uses heat maps to highlight the major image features that the network focused on to make insect species predictions.
6** Badgujar, C.M., Armstrong, P.R., Gerken, A.R., Pordesimo, L.O. and Campbell, J.F. 2023b. Real-time stored product insect detection and identification using deep learning: system integration and extensibility to mobile platforms. J. Stored Prod. Res. 104, 102196. https://doi.org/10.1016/j.jspr.2023.102196 Lasioderma serricorne, Oryzaephilus surinamensis, Sitophilus oryzae, Stegobium paniceum, Tribolium castaneum, Trogoderma variabile. Model often confused and misclassified Lasioderma serricorne and Stegobium paniceum due to their similar appearance. Tribolium castaneum were misidentified due to body having similar shaped to wheat kernels. Accuracy was similar for other species.
Balingbing, C.B.; Kirchner, S.; Siebald, H.; Kaufmann, H.H.; Gummert, M.; Van Hung, N. and Hensel, O. 2024. Application of a multi-layer convolutional neural network model to classify major insect pests in stored rice detected by an acoustic device. Comput. Electron. Agric. 225, 109297. Tribolium castaneum 86.6% Rhyzopertha dominica 85.2%, Sitophilus oryzae 81.8%.
Banga, K.S., Mohapatra, D., Babu, V.B., Giri, S.K., and Bargale, P.C. 2020. Assessment of bruchids density through bioacoustic detection and artificial neural network (ANN) in bulk stored chickpea and green gram. Journal of Stored Products Research, 88: 101667. https://doi.org/10.1016/j.jspr.2020.101667 Artificial neural network (ANN) was applied to predict Callosobruchus chinensis and Callosobruchus maculatus density in bulk stored green gram and chickpea with R2 of 0.99, 0.98 and 0.90, 0.89, respectively.
1 Barboza da Silva, C., Silva, A.A.N., Barroso, G., Yamamoto, P.T., Arthur, V., Toledo, C.F. M., and Mastrangelo, T. de A. 2021. Convolutional neural networks using enhanced radiographs for real-time detection of Sitophilus zeamais in maize grain. Foods 10 (4), 879. MobileNetV2 had the best performance (0.91) on the validation dataset and Xception had best performance on test database (1.00).
27* Bisgin, H., Bera, T., Wu, L., Ding, H., Bisgin, N., Liu, Z., Pava-Ripoll, M., Barnes, A., Campbell, J.F., Vyas, H., Furlanello, C., Tong, W. and Xu, J. 2022. Accurate species identification of food-contaminating beetles with quality-improved elytral images and deep learning. Front. Artif. Intell. 5, 952424. https://doi.org/10.3389/ frai.2022.952424. (Callosobruchus maculatus, Cryptolestes pusillus, Cryptolestes turcicus, Gnathocerus cornutus, Lasioderma serricorne, Latheticus oryzae, Lophocateres pusillus, Oryzaephilus mercator, Oryzaephilus surinamensis, Palorus ratzeburgii, Rhyzopertha dominica, Sitophilus granarius, Sitophilus zeamais, Stegobium paniceum, Tribolium castaneum, Tribolium confusum, Tribolium destructor, Tribolium madens, Trogoderma inclusum >80%), (Ahasverus advena, Araecerus fasciculatus, Attagenus unicolor, Cathartus quadricollis, Cryptolestes ferrugineus, Cynaeus angustus, Sitophilus oryzae, <80%) (Bold only in this study). Explored whether high-quality elytral images can truly achieve near-perfect prediction accuracies for 27 different species using convolutional neural network (CNN) model.
15 Bisgin, H., Bera, T., Ding, H., Semey, H.G., Wu, L., Liu, Z., Barnes, A.E., Langley, D.A., Pava-Ripoll, M., Vyas, H.J., Tong, W. and Xu, J. 2018. Comparing SVM and ANN based machine learning methods for species identification of food contaminating beetles. Sci. Rep. 8, 6532. https://doi.org/10.1038/s41598-018-24926-7. (Ahasverus advena, Cryptolestes pusillus, Gnathocerus cornutus, Lasioderma serricorne, Rhyzopertha dominica, Sitophilus oryzae, Tribolium castaneum, Tribolium madens >80%), (Oryzaephilus mercator, Oryzaephilus surinamensis, Sitophilus granarius, Stegobium paniceum, Tribolium confusum <80%). Support Vector Machine (SVM) model showed slightly better accuracy for most species than Artificial Neural Network (ANN) based pattern recognition in previous study.
1 Boniecki, P., Piekarska-Boniecka, H., ´Swierczynski, K., Koszela, K., Zaborowicz, M., Przybył, J. 2014. Detection of the granary weevil based on X-ray images of damaged wheat kernels. J. Stored Prod. Res. 56, 38–42. https://doi.org/10.1016/j. jspr.2013.11.001 Sitophilua granarius The proposed model identified 100% of the infested kernels correctly, and 98.4% of the healthy ones.
Cao Y, Zhang CJ, Chen QS, Li YY, Qi S, Tian L et al. 2015. Identification of species and geographical strains of Sitophilus oryzae and Sitophilus zeamais using the visible/near-infrared hyperspectral imaging technique. Pest Manag Sci 71:1113–1121 Backpropagation neural network (BPNN) technique classified two species with accuracy of 98%, and distinguish each geographical strain with an accuracy greater than 77.78%.
Casadei, Anita, Maria C. Boukouvala, Gianluca Manduca, Nickolas G. Kavallieratos, Filippo Maggi, Marta Ferrati, Eleonora Spinozzi, Cesare Stefanini, Antonio DeSimone, and Donato Romano 2026. Using deep learning to assess the toxicological effects of sublethal exposure of a novel green pesticide in a stored‐product beetle. Pest Management Science 82(5): 4319-4331.
2** Chen, C., Liang, Y., Zhou, L., Tang, X., and Dai, M. 2022. An automatic inspection system for pest detection in granaries using YOLOv4. Comput. Electron. Agric. 201, 107302. https://doi.org/10.1016/j.compag.2022.107302. Sitophilus oryzae 100% > Tribolium castaneum 95%.
Chen, Long, Rui Li, Chengjun Xie, Peng Chen, Jie Zhang, Jianming Du, Hongbo Chen, Runsheng Qi and Jiahui Sun 2026. A Unified Density-Aware Network for Stored-Grain Pest Detection with Intelligent Risk Assessment. J. Stored Prod. Res. 115, 102838. UDANet risk assessment was investigated over 14 days across grain storage facilities in southern China. It automatically classified and counted insects from images and translated pest counts into actionable risk thresholds (Safe: S ≤ 100, Warning: 100 < S < 300, Emergency: S ≥ 300). Majority of high-risk alerts (83%) were exponentially growing populations of the high-impact species Sitophilus oryzae and Rhyzopertha dominica. Liposcelis bostrychophila is low risk and Oryzaephilus surinamensis, Plodia interpunctella, Ptinus japonica, Trogoderma variabile are moderate risk. Source code is publicly available at: https://github.com/CDragon1/UDANet.
7 Chu, J., Li, Y., Feng, H., Weng, X., Ruan, Y. 2023. Research on multi-scale pest detection and identification method in granary based on improved YOLOv5. Agriculture 13 (2), 364. Bruchidae 99.4% > Sitophilus oryzae 99.3% > Rhyzopertha dominica 99.0% > Plodia interpunctella 98.4% >Tribolium castaneum 98.0% > Cryptolestes ferrugineus 97.3% > Oryzaephilus surinamensis 95.4%.
Deng Z L, Wang P, Song X H, C. Wang, J. Chen, and L. Wu 2020. Research on granary pest detection based on SSD. Computer Engineering and Applications 56(11): 214-218 Sitotroga cerealella 88.3% > Tribolium castaneum 78.6% > Oryzaephilus surinamensis 78.5% > Cryptolestes ferrugineus 75.7% > Rhizopertha dominica 75.3%. The higher accuracy for Sitotroga cerealella using Single Shot MultiBox Detector(SSD) can be attributed to its larger size than other pest species.
Devi, Jyoti, and Anita Singhrova 2026. A Bibliometric Study of IoT and Artificial Intelligence for Grain Storage Management. In 2026 Eight International Conference on Computational Intelligence and Communication Technologies (CCICT), pp. 483-490. BM Institute of Engineering & Technology, DOI: 10.23919/CCICTCPS00010.2026.00085. 211 papers show significant progress in research in this field since 2019 in India, China and the United States.
1 Divyanth, L.G., Chelladurai, V., Loganathan, M., Jayas, D.S., and Soni, P. 2022. Identification of green gram (Vigna radiata) grains infested by Callosobruchus maculatus through X-ray imaging and GAN-based image augmentation. Journal of Biosystems Engineering/ J. Biosyst. Eng. 47 (3), 302–317. https://doi.org/10.1007/s42853-022-00147-9 The overall F1-score produced by the model was improved from 0.86 to 0.91 when the GAN-synthesized dataset additionally supported the training data. Also, the classification accuracy for detecting the stage of internal infestation improved by 5.5%.
Direct-to-satellite IoT sensors 2026. https://lacuna-space.com/remote-grain-monitoring/. Operators receive early alerts on temperature, humidity, and CO₂ changes to stop spoilage before it destroys commercial value.
Du Cong, Wang Yun, Liu Siyu, and Song Xue-hua 2023. Granary pest detection model based on RA-YOLOv5s. Computer Simulation 40(4): 486-491. Rhyzopertha dominica 99. 6%, Oryzaephilus surinamensis 99. 1% > Plodia interpunctella 98. 1%, Sitophilus zeamais 91. 5%, Tribolium castaneum 85. 1% > Cryptolestes ferrugineus 69. 4%. Spatial and channel attention mechanisms added to CSP (Cross Stage Partial) modules of YOLOv5s to facilitate learning of informative features and repeated residual units in the CSP module are removed to make more lightweight.
Erdogan, Pervin, Muhammad Aasim, Zemran Mustafa, Amjad Ali, … and Abdul Waheed 2026. Integrated optimization of Lantana camara-based biopesticide for Sitophilus granarius control: Chemical profiling, residual efficacy, and machine-learning-driven decision support. J. Stored Prod. Res. 119, 103190
Ge, Hongyi, Jing Wang, Tong Zhen, Zhihui Li, Yuhua Zhu, and Quan Pan. 2025. FCA-Yolo: An efficient deep learning framework for real-time monitoring of stored-grain pests in smart warehouses. Agronomy 15(6), 1313. Tribolium castaneum 98.7% > Sitophilus oryzae 96.5% > Cryptolestes ferrugineus 96.1%. FCA-YOLO, based on YOLOv8, achieves 4.51% improvement in accuracy of pest detection. Inspection trolleys were deployed to achieve real-time prediction and analysis of pests in the warehouse.
Gu, Xiangpeng, and Midori Tuda. 2026. Predicting life-history traits in a stored bean pest beetle Callosobruchus chinensis (Coleoptera: Chrysomelidae: Bruchinae) using machine learning. bioRxiv 2026-03
Guan, Zhizhong, Fuyan Sun, Zongwang Lyu, Yiyang Xin, Zihan Zhao 2026. MSGP-DETR: A multi-species stored-grain insect pest detection transformer integrating frequency-domain enhancement and state space models. J. Stored Prod. Res. 116, 102941. Achieving accurate detection through object detection methods remains challenging due to small target scales, subtle inter-species morphological differences, and complex background interference. Rhyzopertha dominica, Tribolium castaneum, Plodia interpunctella, Sitophilus oryzae, Sitotroga cerealella, Oryzaephilus surinamensis, Callosobruchus maculatus, Lasioderma serricorne. MSGP-DETR achieves 94.6 % AP50 accuracy.
Guo, Y X. 2023. Research on target detection algorithm for grain storage pests based on deep learning. Zhengzhou: Henan University of Technology
Guru, P. N., Dhritiman Saha, Yogesh B. Kalnar, Monika Sharma, Ruchika Zalpouri, Virinder Kumar, and Nivedita Shettigar 2025. Predicting phototaxis of almond moth, Cadra cautella (Walker) using ANN models: insights for wavelength and intensity as key factors. Pest Management Science 81(8): 4264-4274. DOI: 10.1002/ps.8787
Haddad, Jamil S., Ibrahim I. Ikhries, Ali F. Al-Shawabkeh, Nijad A. Al-Najdawi, and Fadi A. Alfaqs. 2025. Optimizing the moisture content of sesame seed drying process by enforcing neural network algorithms. Results in Engineering 108789. First study to perform a statistical analysis then applies the artificial neural networks model for the vibratory drying of sesame seeds.
Han, Fei, and Zhijie Leng. 2026. Spatiotemporal prediction model of pest infestation based on deep learning in low temperature grain storage. Journal of Biotech Research 26: 169-180. Model achieved an accuracy of 85.12% for predicting pest risk levels at key locations and 97.56% at the warehouse scale. For 1-, 3-, and 7-day-ahead, prediction rates of high risk grids reached 90.97%, 92.30%, and 83.20%.
HARDWARIO and GrainLink 2026. Revolutionizing Grain Storage with IoT Technology https://www.hardwario.com/cases/hardwario-and-grainlink/. Over 1,000 monitoring systems deployed across Canada and the USA for smart grain-silo monitoring in demanding climatic conditions.
Hou, Li, Mingliang Liu, Yanfen Liu, Xiaodong Tan, Chunxiang Cheng, Wentao Sun, and Nan Tang. 2026. ABF-YOLO11: A multi-scale adaptive detection and lightweight recognition model for stored-grain pests. Journal of Stored Products Research 116, 102932. Rhyzopertha dominica (95.6%) > Tribolium castaneum (94.4%) > Sitophilus granarius (93.2%). Stable performance in complex backgrounds (wheat, corn, paddy), excels at distinguishing similar pest species, and can be seamlessly integrated into smart granary systems (e.g., edge sensors, mobile inspection devices) for real-time, low-power monitoring. ABF-YOLO11 achieves 0.944 mAP50 and 0.923 precision—2.0 % and 4.1 % higher than YOLO11 and 20.6 % lower computational load cost than YOLO11.
Jiang, Xuemei, Yiao Zou, Haojie Li, Ting Guan, Mincheng Bai, Yufeng chen, Lanyue Huang, Hao Feng, Bingzhao Zheng, Zhaolin Gu and Zhongke Qu 2026. Multi-Gas Monitoring System with Enhanced Chemical Resilience for Intelligent Grain Management. J. Stored Prod. Res. 116, 102957.
5 Juan, R.O. Serfa and Gerken, A.R. 2025. Green AI solutions for automated insect monitoring to improve stored product management systems. 2025 IEEE Green Technologies Conference (GreenTech), pp. 143–147. https://doi.org/10.1109/ GreenTech62170.2025.10977676. Tribolium castaneum 99.0% > Oryzaephilus surinamensis 98.8% > Rhyzopertha dominica 98.6% > Sitophilus zeamais 98.5% > Cryptolestes ferrugineus 98.4%.
Junior, W.N.F., Resende, O., de Oliveira, D.C., de Oliveira, D.E.C., and Rosa, E. dos S. 2024. Estimating energy efficiency of the aeration process of stored grains through machine learning. Brazilian J. Agric. Environ. Eng. 28, e281001.
Li, Boyang, Li Liu, Haijiang Jia, Zhaoyang Zang, …, and Jiaqin Xi. 2024a. YOLO-TP: A Lightweight Model for Individual Counting of Lasioderma Serricorne. J. Stored Prod. Res. 109, 102456. High accuracy rate of 99.5%.
Li Dandan, Liu Xinxin, Luo Yuanyi, Liu Qianyuan, Zheng Yili, Chen Jinying, and Zhu Yanguang. 2026a. Research Progress and Challenges in Object Detection of Stored-Grain Pests Based on Deep Learning. Science & Technology of Cereals, Oils & Foods 34(2), 220 (In Chinese with English abstract and captions). One-stage models (YOLO) and two-stage models (Faster R-CNN series, RetinaNet) for deep learning-based detection methods and improvements such as multi-scale feature fusion, frequency-domain feature enhancement, and attention mechanisms are summarized. Application of lightweight networks and model fusion strategies has enhanced real-time performance and deployment potential on mobile and edge devices, driving the transition from laboratory research to practical application. Current challenges are identified, including limitations of available datasets, difficulties in small-object detection, high model complexity, and lack of interpretability. Behavioral characteristics of pests and the temporal dynamics of storage environments remain insufficiently modeled, while research into multimodal fusion and explainable AI is still in its infancy. By constructing open datasets, developing hybrid architecture models, and introducing decision systems driven by behavioral prediction and knowledge graphs, stored-grain pest detection can transition from being merely “algorithm-driven” to “system-enabled,” providing technical support for intelligent grain storage.
Li, Dandan, Xinxin Liu, Jinhui Zhao, Yili Zheng, Fuji Jian, and Xueyan Zhu. 2026b. A new enhanced lightweight detection model to identify stored grain insects on grain bulk surfaces. Engineering Applications of Artificial Intelligence 181, 115551. Sitophilus zeamais 93.9% > Tribolium castaneum 91.9% > Cryptolestes ferrugineus 91.6% > Oryzaephilus surinamensis 91.4% > Rhizopertha dominica 91.3%. Deployment evaluations of YOLO-DInsects model based on the YOLOv11n on edge device confirmed its capability for real-time insect detection under resource-constrained field environments.
6* Li, Dandan, Jida Tian, Jiangtao Li, Muyi Sun, Huiling Zhou, and Yili Zheng 2024b. Establishment of a dataset for detecting pests on the surface of grain bulks. Applied Engineering in Agriculture 40(3): 363-376. Cryptolestes ferrugineus 97.0% = (Sitophilus oryzae, Sitophilus zeamais 97.0%) > Rhizopertha dominica 91.0% > Oryzaephilus surinamensis 88.0% > Tribolium castaneum 85.0%.
3 Li, Jiangtao, Su, Yuwei, Cui, Zhaojun, Tian, Jida, and Zhou, Huiling 2022. A method to establish a synthetic image dataset of stored-product insects for insect detection. IEEE Access 10, 70269–70278. https://doi.org/10.1109/ACCESS.2022.3188282. Tribolium castaneum 93.3% > Sitophilus oryzae 87.9% > Cyptolestes ferrugineus 76.6%.
10 Li, Jiangtao, Zhou, H., Wang, Z., and Jia, Q. 2020. Multi-scale detection of stored-grain insects for intelligent monitoring. Comput. Electron. Agric. 168, 105114. https://doi.org/ 10.1016/j.compag.2019.105114 Lasioderma serricorne 97.0% > (Tribolium castaneum, Tribolium confusum 95.7%) > Oryzaephilus surinamensis 95.1% > (Sitophilus oryzae, Sitophilus zeamais 95.3%) > (Cryptolestes ferrugineus, Cryptolestes pusillus, Cryptolestes turcicus 93.3%) > Rhyzopertha dominica 92.6%. Five feature pyramid levels of the backbone network handled different scales of insects well.
Li, Yane, Jiaqi Song, Lijun Guo, Xiang Weng, and Dalei Song. 2026c. REGAN-BS-YOLOv8: A Novel Multi-Scale Storage Grain Pest Detection Model Established by Integrating Real-ESRGAN and Swin Transformer for YOLOv8. AgriEngineering 8(7): 283. Lasioderma serricorne 99.5% > Alphitobius diaperinus 99.4% > Sitophilus oryzae 98.9% > Plodia interpunctella 98.8% > Bruchidae 98.7% > Cryptolestes ferrugineus 97.4% > Tribolium castaneum 97.2% > Oryzaephilus surinamensis 96.2% > Rhizopertha dominica 93.2%. Swin Transformer and BiFPN are added to YOLOv8n to obtain better perception ability, positioning accuracy, detail retention and edge clarity, improving the recognition accuracy to 97.7%.
10* Li, Zhou, H., Jayas, D.S. and Jia, Q. 2019. Construction of a dataset of stored-grain insects images for intelligent monitoring. Applied Engineering in Agriculture/ Appl. Eng. Agric. 35(4): 647–655. (Sitophilus oryzae, Sitophilus zeamais 96%) > Lasioderma serricorne 94% > (Cryptolestes ferrugineus, Cryptolestes pusillus, Cryptolestes turcicus 92%) = (Tribolium castaneum, Tribolium confusum 92%) > Rhizopertha dominica 90% > Oryzaephilus surinamensis 89%.
Liang J R. 2024. Research on grain pest target detection model and lightweighting system design. Zhengzhou: Henan University of Technology
2 Liu, Zhang, G., Yang, H., Sun, M., Dang, H., Zhou, X., 2018. Application of object detection algorithm in identification of rice weevils and maize weevils. Proceedings of the 2018 2nd International Conference on Deep Learning Technologies, ICDLT ’18, pp. 76–80. https://doi.org/10.1145/3234804.3234820 Sitophilus oryzae 85.6% > Sitophilus zeamais 85.8%
Lu K. 2024. Research and implementation of a pest monitoring system for grain surfaces based on high-definition dome cameras. Beijing: Beijing University of Posts and Telecommunications
Lü, Zongwang, Tiantian Wang, Fuyan Sun and Yuhua Zhu 2025. Improved YOLOv8m- based grain insect detection method for wheat storage. Journal of Chinese Agricultural Mechanization 46(3): 108-114. Sitotroga cerealella 99.5% > Sitophilus zeamais 98.8% Plodia interpunctella 98.7% > Rhyzopertha dominica 93.8% > Tribolium castaneum 86.1%. Replacement of Darknet—53 with Shufflenetv2, a lighter network, in YOLOv8m algorithm, addition Squeeze—and—Excitation Networks attention mechanism at the end of the backbone network to obtain high-quality feature maps, and adopted WIoUv3 Loss function improved accuracy and speed of detection.
Luo, Yuanyi, Dandan Li, Jinying Chen, Yanguang Zhu, Yiming Ma, Jie Lin, Kun Hu et al. 2026. AI-driven technologies for pest monitoring, unsound kernel detection, and intelligent aeration in grain storage. International Journal of Agricultural and Biological Engineering 19(1): 1-10.
5 Lyu, Z., Jin, H., Zhen, T., Sun, F., Xu, H., 2021a. Small object recognition algorithm of grain pests based on SSD feature fusion. IEEE Access 9, 43202–43213. https://doi. org/10.1109/ACCESS.2021.3066510. Weevil 96.9% > Rhyzopertha dominica 95.2% > Tribolium castaneum 94.5% > Cryptolestes ferrugineus 94.0% > Plodia interpunctella 93.0%
Martin, Daniel, Hongjian Ding, Leihong Wu, Howard Semey, Amy Barnes, Darryl Langley, Su Inn Park, Zhichao Liu, Weida Tong and Joshua Xu 2016. An Image Analysis Environment for species indentification for food contaminating beetles. Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence 16: 4375–4376. In collaboration with FDA food analysts, image analysis-based machine intelligence achieved species identification with up to 90% accuracy. This paper is a follow-up to previous research by Park et al. 2016. MATLAB used for image processing and machine learning.
2** Mendoza, Q.A., Pordesimo, L., Neilsen, M., Armstrong, P., Campbell, J. and Mendoza, P.T. 2023. Application of machine learning for insect monitoring in grain facilities. AI 4(1): 348–360. https://doi.org/10.3390/ai4010017. Trogoderma variabile 91.4% > Lasioderma serricorne 90.6%
Montemayor, J.J.M., Escuadra, G.P.G., Nambatac, M.A.G. and Tenoria, D.T. 2024. Detecting rice weevils in stored grains using MFCC and CNN. Proc. Comp. Sci. Seventh Inform. Syst. Int. Conf. (ISICO 2023) 234, 1681–1688. https://doi.org/10.1016/j. procs.2024.03.173. Method for early detection of Rice Weevils, Sitophilus oryzae, based on the frequency of the sound they produce.
Nyabako, Tinashe, Brighton M. Mvumi, Tanya Stathers, Shaw Mlambo, and Macdonald Mubayiwa 2020. Predicting Prostephanus truncatus (Horn) (Coleoptera: Bostrichidae) Populations and Associated Grain Damage in Smallholder Farmers’ Maize Stores: A Machine Learning Approach. J. Stored Prod. Res. 87, 101592. Using decision-tree algorithms and weather-correlated data, the grain damage prediction model achieved a strong correlation of 0.93.
Ogungbite, Olaniyi Charles, and Alaba Bukola Ogungbite 2026. Artificial intelligence in stored-product pest monitoring and decision support: Implications for reduced-risk and botanical control. Journal of Stored Products Research 118, 103074. Artificial Intelligence can strengthe stored-product pest monitoring and management by linking sensing, classification, forecasting, and decision support. AI may improve infestation-risk prediction, spatial targeting, and the timing of reduced-risk interventions. AI is close to closing the long-standing gap between monitoring outputs and useable management decisions.
15* Park, S.I., Bisgin, H., Ding, H., Semey, H.G., Langley, D.A., Tong, W. and Xu, J. 2016. Species identification of food contaminating beetles by recognizing patterns in microscopic images of elytra fragments. PLoS One 11 (6), e0157940. https://doi. org/10.1371/journal.pone.0157940. (Cryptolestes pusillus, Gnathocerus cornutus, Lasioderma serricorne, Sitophilus oryzae, Stegobium paniceum, Tribolium brevicornis, Tribolium madens, Zabrotes subfasciatus >80%), (Oryzaephilus surinamensis, Oryzaephilus mercator, Rhyzopertha dominica, Sitophilus granarius, Tribolium castaneum, Tribolium confusum, Tribolium freemani <80%). Elytra fragments are more commonly recovered from processed food products than other body parts due to their hardness. Local features (feature points of interest identified using keys) are better suited than global features (distinctive shapes, textures, and colors across a whole elytra) for species identification.
4 Qi, R., Li, R., Zhang, J., Xia, Y., Du, J., Sun, J., chen, L., Xie, C., Zhang, H. and Li, G., 2025. A density-point network for dense tiny stored grain pest counting. J. Stored Prod. Res. 111, 102536. https://doi.org/10.1016/j.jspr.2024.102536 Cryptolestes ferrugineus, Liposcelididae, Oryzaephilus surinamensis, Rhyzopertha dominica Our method consistently outperforms others, achieving the lowest MAE and MSE scores across four pest species captured using traps. All methods show high error metrics due to the translucent and small size of booklice, Liposcelididae. Point-based (PB) methods demonstrate relatively good performance by relying on single-point annotations for individual insects, effectively mitigating the issue of overlapping insects.
Ridgway, C., E. R. Davies, J. Chambers, D. R. Mason, and M. Bateman. 2002. AE—Automation and Emerging Technologies: Rapid Machine Vision Method for the Detection of Insects and other Particulate Bio-contaminants of Bulk Grain in Transit. Biosystems Engineering 83(1): 21-30. Recognition success of Oryzaephilus surinamensis adults was 100%.
Rossos, Stavros, Paraskevi Agrafioti, Vasilis Sotiroudas, Christos G. Athanassiou, and Efstathios Kaloudis 2025. Predicting Heat Treatment Duration for Pest Control Using Machine Learning on a Large-Scale Dataset. Agronomy 15(5), 1254. Highlights the importance of tailored heat treatment protocols and the potential of data-driven approaches to optimize pest control strategies, reduce energy consumption, and improve operational efficiency in industrial settings.
6 Shen, Y., Zhou, H., Li, J., Jian, F. and Jayas, D.S. 2018. Detection of stored-grain insects using deep learning. Computers and Electronics in Agriculture/Comput. Electron. Agric. 145, 319–325. https://doi.org/ 10.1016/j.compag.2017.11.039. Tribolium confusum 96.9% > Sitophilus oryzae 95.3% > Rhyzopertha dominica 93.8% > Cryptolestes pusillus 86.2% > Oryzaephilus surinamensis 79.9% >Lasioderma serricorne 72.9%.
8 Shi, Z., Dang, H., Liu, Z., Zhou, X. 2020. Detection and identification of stored-grain insects using deep learning: a more effective neural network. IEEE Access 8, 163703–163714 https://doi.org/10.1109/ACCESS.2020.3021830 Rhyzopertha dominica 93.71% > Lasioderma serricorne 91.05% > Oryzaephilus surinamensis 89.32% > Tribolium confusum 88.94% > Cryptolestes ferrugineus 87.25% > Sitophilus oryzae 86.36% > Sitophilus zeamais 84.63% Cryptolestes turcicus 83.24%. For Sitophilus, average accuracy was 6% lower in simulated grain storage compared to laboratory.
Siaho, Diomande, Pandry Koffi Ghislain, Kadjo Tanon Lambert, Kakou Kouassi Ernest, Oumtanaga Souleymane, and Assidjo Nogbou Emmanuel 2023. Modeling Artificial Neural Network of Insect’s Proliferation During Cocoa Beans Storage. Ingenierie des Systemes d’Information 28(2): 291. Tribolium castaneum, Ephestia cautella, Araecerus faciculatus, Lasioderma serricorne, R²= 0.9982. shows a good correlation of the experimental values and those predicted.
Sun, Fuyan, Zhizhong Guan, Zongwang Lyu, and Shanshan Liu. 2025. High-precision stored-grain insect pest detection method based on PDA-YOLO. Insects 16(6), 610. Sitotroga cerealella 98.6% > Plodia interpunctella 98.1% > Sitophilus zeamais 97.2% > Rhyzopertha dominica 94.7% >Tribolium castaneum 94.5%. PDA-YOLO, an algorithm based on YOLO11n, integrates three key modules to improve stored-grain insect pest detection: PoolFormer_C3k2 (PF_C3k2) for efficient local feature extraction, Attention-based Intra-Scale Feature Interaction (AIFI) for enhanced global context awareness, and Dynamic Multi-scale Aware Edge (DMAE) for precise boundary detection, achieving mAP@0.5 of 96.6%. Demonstrates advantages over mainstream detection algorithms, balancing accuracy, computational efficiency, and real-time performance.
Sun, Fuyan, Zhizhong Guan, Zongwang Lyu, Tong Zhen, Shanshan Liu, and Xiongfei Li. 2026. Efficient and lightweight deep learning model for small-object detection in stored grain pest management. Journal of Stored Products Research 115, 102890. Plodia interpunctella 97.7% = Sitotroga cerealella 97.7% > Rhyzopertha dominica 97.6% > Tribolium castaneum 97.3% > Sitophilus zeamais 97.1%. Pest-YOLO increased precision by 3.9 % compared with 10 published models and was deployed on edge computing platform successfully.
Tadiparthi, Subhash, Chetan M. Badgujar, Alison R. Gerken, Deanna Scheff, P.N. Guru and Sathishkumar Samiappan 2027. Stored-product insect pest detection and identification using deep learning-based computer vision methods: A systematic literature review. J. Stored Prod. Res. 120, 103228. This literature review analyzed 29 studies, only 4 of the 29 studies released public datasets, and only three demonstrated edge or mobile deployment. YOLO and Faster R-CNN variants were the most widely adopted and modified architectures, while recent studies increasingly incorporated attention and transformer-based components, backbone modifications, multi-scale feature extraction, feature pyramid and feature fusion networks to improve small-insect detection. Small insect size, occlusion, overlapping instances, motion blur, illumination variability, dust, cluttered backgrounds, and similarity between insects and grain surfaces continue to affect recognition accuracy.
TeleSense 2026, a California-based company, recently introduced its new monitoring system the TeleSense GrainSafe™. For storage units that do not have existing temperature sensing cables, TeleSense developed a portable solution: the SensorBall™. A number of SensorBalls can be tossed into a grain pile or a horizontal storage (or a truck or a railcar) to collect the data wirelessly. https://millingandgrain.com/wireless-grain-monitoring-using-iot-technology-19532/
6 Tian, J., Sun, M., Zhou, H., Li, J. 2025. PestDet: a unified detection framework for accurate and efficient stored-grain pest detection. Ecol. Inform. 88, 103145. https:// doi.org/10.1016/j.ecoinf.2025.103145. (Sitophilus oryzae, Sitophilus zeamais 96.3%) > Rhizopertha dominica 95.9% > Oryzaephilus surinamensis 93.5% > Cryptolestes ferrugineus 93.3% > Tribolium castaneum 91.3%. Our code and data are available at (https://github.com/IntelligentsystemlabTian/PestDet).
Tuda, M., and Luna-Maldonado, A.I. 2020. Image-based insect species and gender classification by trained supervised machine learning algorithms. Ecol. Inform. 60, 101135. Callosobruchus chinensis, Anisopteromalus calandrae, Heterospilus prosopidis, Average true positive rates (prediction accuracy) of 88.5–98.5% were achieved for within-species sexing of beetles or wasps. For three-species classification and no sexing, the average TP rate reached 99.67–100% by all algorithms.
Xu, Degang,Lu Wang, Fan Li, Yixin Guo and Kuijie Xing 2022. Application of improved Faster RCNN in food insect object detection[J]. Journal of the Chinese Cereals and Oils Association 37(4): 178-186. Sitophilus oryzae 90.77% > Lasioderma serricorne 90.37% = Oryzaephilus surinamensis 90.37% > Tribolium castaneum 89.98% > Cryptolestes ferrugineus 85.46%. Faster R-CNN algorithm was improved by pyramid pooling module handling different scales of insects with white paper or grain background.
Wang Ying. 2024a. Research on small target detection technology for stored-grain pests based on YOLOv5s. Zhengzhou: Henan University of Technology
Wang Zaiqing 2024b. Research and application of target detection algorithm for stored grain insects based on attention mechanism. Zhengzhou: Henan University of Technology
Wu, Jian-Jun, Hao Dang, Miao Li, et al. 2016. The Prediction Research of Population Density Based on Deep Learning in Grain Stored Insects. International Journal of Hybrid Information Technology 9(10): 251–58. Liposcelis entomophila
15 Wu, Leihong, Liu, Z., Bera, T., Ding, H., Langley, D.A., Jenkins-Barnes, A., Furlanello, C., Maggio, V., Tong, W., Xu, J., 2019. A deep learning model to recognize food contaminating beetle species based on elytra fragments. Comput. Electron. Agric. 166, 105002. https://doi.org/10.1016/j.compag.2019.105002. (Cryptolestes pusillus, Gnathocerus cornutus, Oryzaephilus surinamensis, Rhyzopertha dominica, Sitophilus granarius, Stegobium paniceum, Tribolium brevicornis, Tribolium castaneum, Tribolium madens, Zabrotes subfasciatus >80%), (Lasioderma serricorne, Oryzaephilus mercator, Sitophilus oryzae, Tribolium confusum, Tribolium freeman <80%). Automated feature design, transfer learning and deep learning models will help Food and Drug Administration (FDA) investigators quickly and accurately recognize the species of pest in food contamination.
2 Yang, Zhao, H., Zhang, D., Cao, Y., Teng, S.W., Pang, S., Zhou, X., Li, Y., 2022. Auto- identification of two Sitophilus sibling species on stored wheat using deep convolutional neural network. Pest Manag. Sci. 78 (5), 1925–1937. https://doi.org/ 10.1002/ps.6810. Sitophilus oryzae 92.6% > Sitophilus zeamais 88.1%. S. zeamais adults are darker and shinier than S. oryzae in visible light. Upgraded industrial camera system have been installed in over 100 000 grain depots of China.
Yu, Jingxiao, Hongbin Pu, and Da-Wen Sun. 2025. Fast real-time monitor of rice grains infested with Sitophilus oryzae based on terahertz imaging combined with machine learning. Food Control 176, 111290. The enhanced support vector machine classification (SVC) model achieved 100 % accuracy for Sitophilus oryzae while reducing the predictive time. Gradient-weighted class activation mapping (Grad-CAM) indicates attention-assisted VGG7 model can mine important information in the image more accurately and quickly.
Yu, Shuxuan, Lizhuotong Xu, Hong Cheng, Chen Tian, Qianyu Hang, Siqi Zhao, Peian Tang, Chao Ding, and Qiang Liu. 2026. Integrating texture features with hierarchical machine learning classifiers for few shot stored grain pest identification. Journal of Stored Products Research 119, 103115. Sitophilus zeamais 96% > Cryptolestes pusillus 95% > Tribolium castaneum 94% = Tribolium confusum 94% > Rhyzopertha dominica 93% > Oryzaephilus surinamensis 91% > Cryptolestes ferrugineus 89%. Evaluates body surfaces texture features to enhance classification accuracy.
Zayas I Y and Flinn P W. 1998. Detection of insects in bulk wheat samples with machine vision. Transactions of the ASAE 41: 883-888. There is no published research on detection of insects in the bulk grain using digital imaging technique. Recognition success of Rhyzopertha dominica adults, some grass seeds, and other non-wheat components was higher than 90%.
Zhang, Chunlei 2024a. Research on granary pest detection device and pest identification algorithm. Hangzhou: Zhejiang A & F University
Zhang, Hongwei, Jingyu Yang, Yating Zhu, Zhuo Gao, Enze Ren, and Bixian Li. 2026. A Thermal State Field Forecasting Framework for Intelligent Ventilation Decision-Making in Granary Based on a Hierarchical TCN-Mamba Network. Sensors 26(17): 5362. (Flat storage)
Zhang, Hongwei, Jingyu Yang, Yating Zhu, Enze Ren, … and Bixian Li 2027. A multi-feature prediction method for grain pile temperature based on regional modeling and deep spatio-temporal learning. J. Stored Prod. Res. 120, 103256.
Zhang, Xiaoyu, Jiale Guo, Yanhong Liu, Yuanyi Luo, … and Xiaoping Yan 2026. Intelligent aeration technologies in grain storage: A comprehensive review of multi-physics modeling and AI-driven control. J. Stored Prod. Res. 119, 103155.
6 Zhang, Yingying, Zhong, Weibo and Pan, Hu 2021. Identification of stored grain pests by modified residual network. Comput. Electron. Agric. 182, 105983. https://doi.org/10.1016/j. compag.2021.105983. Sitotroga cerealella 100% > Cryptolestes pusillus 98% > Rhizopertha dominica 97% > Oryzaephilus surinamensis 95% > Sitophilus oryzae 95% = Sitophilus zeamais 95%
Zhang Zhuang, Fu Hui, Wang Dian Xuan, Jing Shihua, Li Zhixin, Wang Ze and Liu Man 2024b. Application of AI multi-model technology in prevention and control of stored grain pests. Journal of the Chinese Cereals and Oils Association 39(11): 17-25. maize weevil, adzuki bean weevil, lesser grain borer, various grain beetles, and grain moths. Gaussian Mixture Model (GMM) is employed to capture moving pest images from video and three-frame differencing method generates a complete binary image of the pests which is identified by ResNet18 architecture. Model predicts need and estimates efficacy of insect control by fumigation or controlled-atmosphere. Grain loss caused by pests during storage at the Jinan No. 3 Grain Depot was reduced by 56% compared to previous levels, and storage costs were lowered by 35%.
Zhao, Chao, Songbo Li, Duangsamorn Suthisut, Chunqi Bai, Lei Yan, Dianxuan Wang, Jianhua Lü, and Peng Li. 2026. OS-YOLO: A robust object detection model for storage pest monitoring in American ginseng. Journal of Agriculture and Food Research 26, 102751. Plodia interpunctella 98.3% > Tribolium castaneum 95.6% > Sitophilus zeamais 95.3% > Stegobium paniceum 95.1% > Oryzaephilus surinamensis 89.2%. OS-YOLO achieved a precision of 96.1% outperforming several state-of-the-art detection models for five major insect pests of Chinese medical material.
12 Zhao, C., Bai, C., Yan, L., Xiong, H., Suthisut, D., Pobsuk, P. and Wang, D. 2024a. AC-YOLO: multi-category and high-precision detection model for stored grain pests based on integrated multiple attention mechanisms. Expert Syst. Appl. 255, 124659. https:// doi.org/10.1016/j.eswa.2024.124659. Plodia interpunctella 98.3% > Cynaeus angustus 98.2% > Sitophilus zeamais 97.4% > Gibbium aequinoctiale 97.3% > Tribolium castaneum 96.4% > Ephestia elutella 92.4% > Oryzaephilus surinamensis 91.0% > Cryptolestes ferrugineus 90.0% > Rhyzopertha dominica 89.1% > Cryptolestes pusillus 86.3% > Lasioderma serricorne 84.8% > Sitotroga cerealella 81.8% (Bold only in this study). Similarity in color between wheat and Sitotroga cerealella makes accurate detection challenging. Convolutional Block Attention Module (CBAM) was integrated into the backbone network of YOLOv7 algorithm to improve the perception of crucial features, and convolutional self-attention mechanism (ACmix module) was added to boost the feature perception by reducing attention to interfering backgrounds, thereby enhancing detection accuracy with improvement of 6.2 percentage points. Most significant improvements are observed with 15.1 percentage points increase for S. cerealella, 12.2 for C. ferrugineus, 11.2 for O. surinamensis, 7.5 for E. elutella and 7.1 for P. interpunctella.
5 Zhou, Miao, H., Li, J., Jian, F., Jayas, D.S., 2019. A low-resolution image restoration classifier network to identify stored-grain insects from images of sticky boards. Comput. Electron. Agric. 162, 593–601. https://doi.org/10.1016/j. compag.2019.05.015. Lasioderma serricorne, Oryzaephilus surinamensis, Rhizopertha dominica, Sitophilus oryzae, Tribolium castaneum
6 Zhu, Li, D., Zheng, Y., Ma, Y., Yan, X., Zhou, Q., Wang, Q., Zheng, Y. 2025. A YOLO- based model for detecting stored-grain insects on surface of grain bulks. Insects 16 (2), 210. Cryptolestes ferrugineus, Oryzaephilus surinamensis, Rhyzopertha dominica, Sitophilus zeamais, Sitophilus oryzae, Tribolium castaneum. Similarities of shape and color of T. castaneum and R. dominica make them more difficult to distinguish from one another.
