In recent years, global agriculture has encountered several challenges exacerbated by the effects of changes in climate, such as extreme water shortages for irrigation and heat waves. Water-deficit stress adversely affects the morpho-physiology of numerous crops, including soybean (Glycine max L.), which is considered as promising crop in Bangladesh. Seaweed extract (SWE) has the potential to improve crop yield and alleviate the adverse effects of water-deficit stress. Remote and proximal sensing are also extensively utilized in estimating morpho-physiological traits owing to their cost-efficiency and non-destructive characteristics. The study was carried out to evaluate soybean morpho-physiological traits under the application of water extracts of Gracilaria tenuistipitata var. liui (red seaweed) with two varying irrigation water conditions (100% of total crop water requirement (TCWR) and 70% of TCWR). Principal component analysis (PCA) revealed that among the four treatments, the 70% irrigation + 5% (v/v) SWE and the 100% irrigation treatments overlapped, indicating that the application of SWE effectively mitigated water-deficit stress in soybeans. This result demonstrates that the foliar application of 5% SWE enabled soybeans to achieve morpho-physiological performance comparable to that of fully irrigated plants while reducing irrigation water use by 30%. Based on Pearson’s correlation matrix, a simple linear regression model was used to ascertain the relationship between unmanned aerial vehicle (UAV)-derived vegetation indices and the field-measured physiological characteristics of soybean. The Normalized Difference Red Edge (NDRE) strongly correlated with stomatal conductance (R² = 0.76), photosystem II efficiency (R² = 0.78), maximum fluorescence (R² = 0.64), and apparent transpiration rate (R² = 0.69). The Soil Adjusted Vegetation Index (SAVI) had the highest correlation with leaf relative water content (R² = 0.87), the Blue Normalized Difference Vegetation Index (bNDVI) with steady-state fluorescence (R² = 0.56) and vapor pressure deficit (R² = 0.74), and the Green Normalized Difference Vegetation Index (gNDVI) with chlorophyll content (R² = 0.73). Our results demonstrate how UAV and physiological data can be integrated to improve precision soybean farming and support sustainable soybean production under water-deficit stress.

Despite progress in AI-based plant diagnostics, sugarcane farmers in low-resource regions remain vulnerable to leaf diseases due to the lack of scalable, efficient, and interpretable tools. Many deep learning models fail to generalize under real-world conditions and require substantial computational resources, limiting their use in resource-constrained regions. In this paper, we present SugarcaneLD-BD, a curated dataset for sugarcane leaf-disease classification; SugarcaneShuffleNet, an optimized lightweight model for rapid on-device diagnosis; and SugarcaneAI, a Progressive Web Application for field deployment. SugarcaneLD-BD contains 638 curated images across five classes, including four major sugarcane diseases, collected in Bangladesh under diverse field conditions and verified by expert pathologists. To enhance diversity, we combined SugarcaneLD-BD with two additional datasets, yielding a larger and more representative corpus. Our optimized model, SugarcaneShuffleNet, offers the best trade-off between speed and accuracy for real-time, on-device diagnosis. This 9.26 MB model achieved 98.02% accuracy, an F1-score of 0.98, and an average inference time of 4.14 ms per image. For comparison, we fine-tuned five other lightweight convolutional neural networks: MnasNet, EdgeNeXt, EfficientNet-Lite, MobileNet, and SqueezeNet via transfer learning and Bayesian optimization. MnasNet and EdgeNeXt achieved comparable accuracy to SugarcaneShuffleNet, but required significantly more parameters, memory, and computation, limiting their suitability for low-resource deployment. We integrate SugarcaneShuffleNet into SugarcaneAI, delivering Grad-CAM-based explanations in the field. Together, these contributions offer a diverse benchmark, efficient models for low-resource environments, and a practical tool for sugarcane disease classification. It spans varied lighting, backgrounds and devices used on-farm

Soybean improvement depends on the identification of genetically diverse germplasm with superior agronomic performance. This study evaluated the phenotypic diversity, yield potential, and genetic variability of a large and diverse panel of 278 soybean genotypes under the agro-ecological conditions of the Madhupur Tract, Bangladesh, during November 2023 to April 2024. The experiment was conducted in a randomized complete block design with two replications. Morphological, phenological, physiological, and yield-related traits were recorded and analyzed using analysis of variance, correlation, principal component analysis (PCA), and genetic parameter estimation. Considerable variation was observed for all evaluated traits. Pointed ovate leaves (58.3%), white flowers (38.1%), indeterminate growth habit (97.8%), medium plant height (64.7%), medium maturity (85.3%), and medium seed size (54.3%) predominated among the genotypes. Grain yield per plant (GY) ranged from 0.46 to 20.06 g, with BS-53 recording the highest yield, while several other genotypes also exhibited comparatively superior yield and favorable agronomic characteristics. BS-260 matured earliest (88 days) and BS-18 exhibited the highest leaf chlorophyll content. GY showed strong positive correlations with seeds per plant (r = 0.906 ***) and pods per plant (r = 0.862 ***). PCA explained 67.4% of the total phenotypic variation in the first two principal components. High broad-sense heritability coupled with high genetic advance for most traits suggested substantial genetic variability and potential responsiveness to selection under the conditions of the present study. The comprehensive characterization of this diverse germplasm panel and the identification of high-yielding and agronomically favorable genotypes provide a valuable resource for soybean selection and breeding in Bangladesh, while the promising genotypes warrant further multi-location and multi-season validation.