Novel hybrid transfer neural network for wheat crop growth stages recognition using field images

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Апрель 7, 2025

Wheat is one of the world's most widely cultivated cereal crops and a primary food source for significant portion population. goes through several distinct developmental phases, accurately identifying these stages essential precision farming. Determining wheat growth crucial increasing efficiency agricultural yield in Preliminary research identified obstacles distinguishing between stages, negatively impacting crop yields. To address this, this study introduces an innovative approach, MobDenNet, based on data collection real-time stage recognition. The utilized diverse image dataset covering seven phases 'Crown Root', 'Tillering', 'Mid Vegetative', 'Booting', 'Heading', 'Anthesis', 'Milking', comprising 4496 images. collected underwent rigorous preprocessing advanced augmentation to refine minimize biases. This employed deep transfer learning models, including MobileNetV2, DenseNet-121, NASNet-Large, InceptionV3, convolutional neural network (CNN) performance comparison. Experimental evaluations demonstrated that model MobileNetV2 achieved 95% accuracy, DenseNet-121 94% NASNet-Large 76% InceptionV3 74% CNN 68% accuracy. proposed novel hybrid synergistically merges architectures networks, yields highly accurate results with precision, recall, F1 score 99%. We validated robustness approach using k-fold cross-validation. ensures detection great promise boosting productivity management practices, empowering farmers optimize resource distribution make informed decisions.

Язык: Английский

---

The Neural Frontier of Future Medical Imaging: A Review of Deep Learning for Brain Tumor Detection

Journal of Imaging, Год журнала: 2024, Номер 11(1), С. 2 - 2

Опубликована: Дек. 24, 2024

Brain tumor detection is crucial in medical research due to high mortality rates and treatment challenges. Early accurate diagnosis vital for improving patient outcomes, however, traditional methods, such as manual Magnetic Resonance Imaging (MRI) analysis, are often time-consuming error-prone. The rise of deep learning has led advanced models automated brain feature extraction, segmentation, classification. Despite these advancements, comprehensive reviews synthesizing recent findings remain scarce. By analyzing over 100 papers past half-decade (2019-2024), this review fills that gap, exploring the latest methods paradigms, summarizing key concepts, challenges, datasets, offering insights into future directions using learning. This also incorporates an analysis previous targets three main aspects: results revealed primarily focuses on Convolutional Neural Networks (CNNs) their variants, with a strong emphasis transfer pre-trained models. Other Generative Adversarial (GANs) Autoencoders, used while Recurrent (RNNs) employed time-sequence modeling. Some integrate Internet Things (IoT) frameworks or federated real-time diagnostics privacy, paired optimization algorithms. However, adoption eXplainable AI (XAI) remains limited, despite its importance building trust diagnostics. Finally, outlines opportunities, focusing image quality, underexplored techniques, expanding deeper representations model behavior recurrent expansion advance imaging

Язык: Английский

---

Enhancing brain tumor detection: a novel CNN approach with advanced activation functions for accurate medical imaging analysis

Frontiers in Oncology, Год журнала: 2024, Номер 14

Опубликована: Сен. 20, 2024

Introduction Brain tumors are characterized by abnormal cell growth within or around the brain, posing severe health risks often associated with high mortality rates. Various imaging techniques, including magnetic resonance (MRI), commonly employed to visualize brain and identify malignant growths. Computer-aided diagnosis tools (CAD) utilizing Convolutional Neural Networks (CNNs) have proven effective in feature extraction predictive analysis across diverse medical modalities. Methods This study explores a CNN trained evaluated nine activation functions, encompassing eight established ones from literature modified version of soft sign function. Results The latter demonstrates notable efficacy discriminating between four types MR images, achieving an accuracy 97.6%. sensitivity for glioma is 93.7%; meningioma, it 97.4%; cases no tumor, 98.8%; pituitary tumors, reaches 100%. Discussion In this manuscript, we propose advanced architecture that integrates newly developed Our extensive experimentation showcase model's remarkable ability precisely distinguish different substantial dataset. findings our suggest model could serve as invaluable supplementary tool healthcare practitioners, specialized professionals resident physicians, accurate tumors.

Язык: Английский

---

MRI brain tumor classification based on CNN features and machine learning classifiers

Journal of Ambient Intelligence and Humanized Computing, Год журнала: 2024, Номер 16(1), С. 233 - 242

Опубликована: Дек. 14, 2024

Язык: Английский

---