UAV swarm passive positioning technology only requires the reception of electromagnetic signals to achieve the positioning and tracking of radiation sources. It avoids the active positioning strategy that requires active emission of signals and has the advantages of good concealment, long acting distance, and strong anti-interference ability, which has received more and more attention. In this paper, we propose a new UAV swarm formation flight method based on pure azimuth passive positioning. Specifically, we propose a two-circle positioning model, which describes the positional deviation of the receiving UAV using trigonometric functions relative to the target in polar coordinates. Furthermore, we design a two-step adjustment strategy that enables the receiving UAV to reach the target position efficiently. Based on the above design, we constructed an optimized UAV swarm formation scheme. In experiments with UAV numbers of 8 and 20, compared to the representative comparison strategy, the proposed UAV formation scheme reduces the total length of the UAV formation by 34.76% and 55.34%, respectively. It demonstrates the effectiveness of the proposed method in the application of assigning target positions to UAV swarms.

Olive oil adulteration with low-cost vegetable oils poses a serious food safety concern. This study proposes a Dynamic Task Priority Multi-Gate Mixture-of-Experts (DTP-MMoE) model based on Raman spectroscopy to simultaneously perform the qualitative discrimination of adulteration types and quantitative prediction of adulteration ratios. The model learns shared spectral representations through expert networks and task-specific gating mechanisms, while a dynamic task priority loss function adaptively balances optimization between the classification and regression tasks. Experimental results demonstrated that the DTP-MMoE model achieved a classification accuracy of 99.15% and a coefficient of determination (R²) of 0.99 for prediction, significantly outperforming conventional single-task and multi-task baselines. Ablation studies confirmed the critical contributions of the gating mechanism, expert network configuration, and dynamic weighting strategy. Furthermore, external validation on commercial blended oil samples not involved in training yielded a mean absolute error (MAE) of 0.317%, an RMSE of 0.459%, and a MAPE of 6.34%, demonstrating good generalization capability. The proposed method provides an efficient, non-destructive, and reliable analytical tool for rapid screening of olive oil authenticity, showing considerable promise for application in food quality control and regulatory practice.

Artificial intelligence-assisted food detection is increasingly moving from single-task classification toward integrated analytical systems capable of producing multiple quality-related outputs from one sensing workflow. However, most hyperspectral food detection studies still rely on a single spectral range or simple feature concatenation, which limits their ability to exploit complementary physicochemical information from heterogeneous sensors. In this study, an artificial intelligence-enabled visible–near-infrared and short-wave infrared (VNIR-SWIR) hyperspectral fusion framework is proposed for multi-task fruit detection, using origin authentication and bruise localization as representative tasks. The proposed method first constructs an observation-consistent fused representation from high-resolution VNIR images and low-resolution SWIR images. Collaborative spectral unmixing is used to couple cross-modal material distributions, while abundance-consistency and downsampled observation-consistency constraints are introduced to estimate SWIR-informed features on the VNIR spatial grid without assuming measured high-resolution SWIR ground truth. The fused representation is then processed by a shared spectral–spatial deep encoder with two task-specific heads: a fruit-level classification head for origin authentication and a pixel-level segmentation head for bruise detection. Experiments on apple and kiwifruit datasets show that the proposed framework outperforms VNIR-only, SWIR-only, bicubic-fusion, CNMF-style fusion, and TV-regularized fusion baselines under five fruit-level stratified random splits. For origin-category authentication, the proposed method achieved an accuracy of almost 93.85 for apples and almost 94.35 for kiwifruit. For bruise localization, the proposed method achieved higher overall accuracy, average accuracy, and Cohen’s kappa across the evaluated fruit categories.

Aflatoxin B1 (AFB1) contamination in peanuts remains a major food-safety concern. Hyperspectralimaging provides continuous spectral signatures linked to sample chemistry and supportsnondestructive screening. However, pixel-level inspection is computationally expensive and difficult todeploy online. Object-level mean spectra are more efficient, but early AFB1 threshold-exceedancesamples show weak and overlapping spectral signatures. A vision-language framework, SpecVLM,was developed to address this limitation. Region-of-interest mean spectra from individual kernelswere corrected by standard normal variate transformation and converted into scale-wavelengthspectrograms by continuous wavelet transform. The spectrograms were paired with task-aware promptsfor supervised multimodal fine-tuning. A spectral detail residual adapter was introduced to enhanceweak spectral cues. Low-rank adaptation was then combined with the adapter on Qwen3-VL-2B forbinary classification. Compared with raw spectra, CWT reduced feature overlap. SpecVLM achieved94.65% accuracy, 93.93% precision, 94.25% recall, and 94.08% F1-score. Ablation tests confirmedthe contribution of the adapter. Grad-CAM responses were concentrated near mold- andcomponentsensitive bands. Therefore, SpecVLM provides an object-level spectrogram-text supervised strategyfor rapid nondestructive screening of AFB1 threshold-exceedance risk in peanuts

The core volume ratio (CVR) is a key indicator for evaluating the proportion of edible fraction in stone fruits. Traditionally, CVR is determined through destructive sampling by separately measuring the masses of the core and entire fruit. Recently, low-field nuclear magnetic resonance imaging (LF-NMRI) has been introduced as a non-destructive alternative, but its sparse sampling limits the ability to achieve accurate spatial and volumetric quantification of fruit quality. To address this limitation, we propose a novel method for high-precision three-dimensional (3D) modeling of stone fruits. The method acquires tomographic LF-NMRI sequences along three orthogonal axes. Each sequence is segmented into pulp and core regions using a SwinUNet deep learning model and converted into point clouds for each view. Point clouds from the three orthogonal views are registered via a genetic algorithm to align structural information from complementary perspectives and fused into a unified 3D model through Poisson surface reconstruction. Using prunes as a representative case, the method enables accurate quantification of core and entire fruit volumes, achieving a CVR estimation with a mean absolute error of 0.13% compared to manual measurements. The proposed three-view reconstruction strategy yields a volumetric error of only 0.73%, significantly outperforming single-view (4.57%) and dual-view (3.73%) approaches. This technology provides a robust and accurate non-destructive solution for 3D internal quality analysis of fruits.