The preservation of commercial value and economic viability is dependent on the proper handling of a mango's post-harvest quality. Vision-based automated defect classification is scalable compared to manual inspection. But previous approaches have had issues with generalizability, interpretability, and using stand alone CNN models. This research tackles these issues with a jointly-distributed, mixed explainability deep learning model, built using complementary deep features of ResNet50 and MobileNetV2, which employs PCA followed by a Gradient Boosting Classifier (GBC) to classify features. The model was tested on a collection of 1200 images of mangoes collected from four characteristics (healthy, black-spotted, shriveled, infected) from various tiered mixed-light orchards and under a range of lighting. The tested system fusion-based architecture enhances performance further with 95.7% accuracy with single model baselines and CNN + SVM combinations with 3–5% gap and overall enhancement of precision, recall, F1 score and MCC. Further analyses using Grad-CAM helped to center and depict the specific areas with the defects. This research clearly demonstrates the power of the feature fusion approach coupled with the ensemble learning and the explainable AI techniques in building the working system for the automated quality assessment of the fruit.
ArivazhaganAShebiahRNAnanthiS, et al.Detection of unhealthy region of plant leaves and classification of plant leaf diseases using texture features. Agric Eng Int: CIGR J2013; 15: 211–217.
2.
BrahimiMBoukhalfaKMoussaouiAM. Deep learning for tomato diseases: classification and symptoms visualization. Appl Artif Intell2017; 31: 299–315.
3.
JadhavJBodheS. Deep learning based system for detection of peel defects in mangoes. Procedia Comput Sci2020; 167: 2191–2198.
4.
GhosalA, et al.CNN-XGBoost: a deep learning framework for weed detection in crops. Comput Electron Agric2020; 172: 105342.
5.
MakmudSTHaryantoANPrasetyoSY. Improving rotten fruit classification accuracy through fusion of multiple pretrained CNN models. Engineering and Applied Science Research2024; 51: 313–320.
6.
HashmiHDwivediRKKumarA. Identification of objects using AI & ML approaches: state-of-the-art. In: 2021 10th International Conference on System Modeling & Advancement in Research Trends (SMART), 2021, pp.1–5. https://doi.org/10.1109/SMART52563.2021.9676273.
7.
SandlerMHowardAZhuM, et al.Mobilenetv2: inverted residuals and linear bottlenecks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018, pp.4510–4520.
8.
ChenTGuestrinC. XGBoost: a scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016, pp.785–794.
9.
KumarAHashmiHKhanSA, et al.SSE: a smart framework for live video streaming-based alerting system. In: 2021 10th International Conference on System Modeling & Advancement in Research Trends (SMART), 2021, pp. 193–197. https://doi.org/10.1109/SMART52563.2021.9675306.
10.
RahmanMT, et al.Identification and classification of mango fruit diseases using convolutional neural network. In: International Conference on Electrical, Computer and Communication Engineering (ECCE), 2019, pp.1–5.
11.
HughesDPSalathéM. An open access repository of images on plant health to enable the development of mobile disease diagnostics. arXiv Preprint, arXiv:1511.08060. 2015.
12.
HashmiHDwivediRKKumarA. YOLO-RS: an efficient YOLO-based approach for remote sensing object detection. In: 2023 12th International Conference on System Modeling & Advancement in Research Trends (SMART), 2023, pp.50–56). https://doi.org/10.1109/SMART59791.2023.10428504.
FuentesJYoonSKimS, et al.A robust deep-learning detector for real-time tomato plant diseases and pests recognition. Sensors2017; 17: 2022.
15.
TooJYujianLNjukiS, et al.A comparative study of fine-tuning deep learning models for plant disease identification. Comput Electron Agric2019; 161: 272–279.
16.
BhosaleAShindeMKulkarniR. Visual explanation of deep learning predictions in plant disease classification using Grad-CAM. Int J Eng Res Technol2022; 11: 320–325.
17.
KamilarisRPrenafeta-BoldúF. Deep learning in agriculture: a survey. Comput Electron Agric2018; 147: 70–90.
18.
HashmiHKumar DwivediRKumarA. Comparative analysis of CNN-based smart pre-trained models for object detection on DOTA. J Autom Mobile Rob Intell Syst2024; 18: 31–45.
19.
LakshmiBHussainMI. Plant leaf disease detection using deep learning and convolutional neural networks. Int J Eng Technol2018; 7: 519–522.
20.
HashmiHDwivediRKumarA. An AI & ML based detection & identification in remote imagery: state-of-the-art. J Autom Mobile Rob Intell Syst2022; 15: 3–17.
21.
KaurSGandhiVGuptaM. Deep convolutional neural networks for mango leaf disease detection. Int J Adv Res Comput Sci2018; 9: 325–328.
22.
MohantySHughesDPSalathéM. Using deep learning for image-based plant disease detection. Front Plant Sci2016; 7: 1419.
23.
ZhangTZhengCQinY. Fusion of deep features for plant disease recognition. IEEE Access2019; 7: 172877–172885.
24.
KumarAHashmiHSinghP, et al.Intelligent surveillance framework for crowd detection and alerting. In: SNSFAIT 2024: International Symposium on Securing Next-Generation Systems using Future Artificial Intelligence Technologies (Vol. 3774). CEUR Workshop Proceedings, 2024.
25.
NazirNFerozeAKhaliqA, et al.Fusion of convolutional neural network and traditional features for fruit classification. SN Appl Sci2019; 1: 1–10.
26.
BrahimiNBoukhalfaKMoussaouiM. Deep learning for plant disease detection and diagnosis: a review. Comput Electron Agric2020; 178: 105790.
27.
ZhangLWangQLiH, et al.A lightweight convolutional neural network for real-field crop disease detection under varying illumination. Comput Electron Agric2023; 210: 107806.
28.
YangBLiMWangY, et al.Plant type, leaf disease and severity identification using CNN and transformer with a multi-label framework. Scientific Reports2024; 14: 11664. DOI: 10.1038/s41598-024-62452-x
29.
ChiouKDTufailADe Silva LiyanageC, et al.. Application of deep learning for fruit defect recognition in Psidium guajava L. Scientific Reports2025; 15. https://doi.org/10.1038/s41598-025-88936-y
30.
DasAPalMChattopadhyayR. Tea leaf disease recognition using fusion of handcrafted and CNN features. Multimed Tools Appl2022; 81: 9697–9716.
31.
BarmanPMitraSBanerjeeS. Multi-stream deep feature fusion for multi-class fruit defect detection. Eng Appl Artif Intell2023; 125: 107007.
SladojevicSArsenovicMAnderlaA, et al.Deep neural networks based recognition of plant diseases by leaf image classification. Comput Intell Neurosci2016; 2016: 1–11.
36.
MishraPSharmaSJoshiR. Tomato leaf disease detection using hybrid deep learning approach. In: International Conference on Computational Intelligence and Smart Communication (CISC), 2021, pp.210–215.
37.
GhoseAMandalSSenapatiS. LightGBM on deep CNN features for pest severity scoring. Neural Process Lett2023; 55: 3421–3438.
38.
RahmanSChowdhuryMAlamT. Gradient boosting classifier over PCA-compressed CNN features for tomato ripeness classification. Pattern Recognit Lett2024; 175: 67–75.
39.
VelayuthamSPremkumarJRajkumarA. SVM Vs. gradient boosting for hyperspectral mango lesion detection. Infrared Phys Technol2025; 138: 105989.
40.
SelvarajuRR, et al.Grad-CAM: visual explanations from deep networks via gradient-based localization. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2017, pp.618–626.
41.
AtilaUInceMCAksoyS. SHAP-Based interpretability for deep citrus defect classification. Comput Electron Agric2023; 205: 107690.
42.
OmarMEl-HakimASheikhH. Grad-CAM and LIME fusion for explainable apple disease localization. J Imaging2024; 10: 55.
43.
ChenYXuLGuoQ. ScoreCAM-enhanced saliency mapping for early crop stress identification. IEEE Trans Artif Intell2025; 6: 43–56.
44.
BhujbalAPatilVWaniR. Mango disease detection and classification using image processing techniques. Int J Innovative Res Sci Eng Technol2022; 11: 2347–2353.
45.
UllahFAhmadJKimDH. Multi-objective resource optimization in UAV-enabled heterogeneous cellular networks using serverless federated learning and power-domain NOMA. IEEE Access2023; 11: 55678–55692.
46.
SuoDLiGXuH. Dynamic offloading in mobile edge computing with traffic-aware network slicing and adaptive TD3 strategy. Comput Netw2024; 242: 110000.
47.
ZhouTLiWZhaoX. Reliable information delivery and dynamic link utilization in MANET cloud using deep reinforcement learning. J Netw Comput Appl2023; 224: 103800.
48.
ShafikWTufailADe Silva LiyanageC, et al.Using transfer learning-based plant disease classification and detection for sustainable agriculture.
49.
PacalIKunduraciogluIAlmaMH, et al. A systematic review of deep learning techniques for plant diseases. Artificial Intelligence Review2024; 57: 304.
50.
DubeySRJalalA. Detection and classification of apple fruit diseases using complete local binary patterns. In: Proceedings of the 3rd International Conference on Computer and Communication Technology (ICCCT), 2012, pp.346–351. https://doi.org/10.1109/ICCCT.2012.74.
51.
AmaraJBouazizBAlgergawyA. A deep learning-based approach for banana leaf diseases classification. In: BTW Workshop on Big Data in Agriculture, 2017, pp.79–88.
52.
SinghAMisraR. Detection of potato plant diseases using CNN-SVM hybrid model. Int J Sci Res Comput Sci Eng Inf Technol2021; 6: 2456–3307.
53.
HeKZhangXRenS, et al.Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp.770–778.
