Search

Accurate agricultural yield estimates are vital for effective food resource management, and the efficacy of yield remote-sensing models is influenced by factors such as geographic location and temporal variations, posing challenges to both the accuracy and transferability. This study used multispectral (MS), thermal infrared (TIR), red-green-blue (RGB) sensors, and LiDAR, alongside soil and precipitation data, over three locations for two years, resulting in six datasets. Three models, stacking ensemble learning (SEL), long short-term memory (LSTM), and LSTM-multi-head self-attention (LSTM-MH-SA), were compared for wheat yield estimation. The flowering stage showed the highest accuracy with SEL ( $R^{2} = 0.83$ ), but the LSTM-MH-SA model outperformed SEL when integrating multitemporal features. As more data types were introduced, precision improved. Specifically, the LSTM-MH-SA model using vegetation indices, texture features (TFs), geographical information, and climate data showed superior accuracy, enhancing $R^{2}$ by 29.58% and reducing errors compared to a basic vegetation indices model. Incorporating the joint distribution adaptation (JDA) method enabled model transfer between varying datasets, maintaining a stable accuracy ( $R^{2} = 0.81$ ) with just 4% target dataset supplementation. Combining deep learning (DL) with transfer learning provides an innovative method for more efficient agricultural yield prediction.

This article deals with the problem of tracking targets with X-band marine radars in the complicated sea clutter background. By jointly exploiting the target’s kinematic and appearance information, we propose a novel dynamic model-based Doppler-adaptive correlation filter (DDACF). The proposed tracker mainly consists of two modules. First, a DACF is developed based on the kernel correlation filter. This filter can effectively represent the appearance patterns of the target and is adaptive to the motion Doppler by using a multifrequency-centered filter bank. Second, the Bernoulli filter is employed to represent the kinematic patterns of the target. Within this filter, the converted measurement Kalman filter with range rate (CMKFRR) is applied to overcome the inconsistency between coordinate systems of the motion and measurement models. By exploiting the hybrid measurement likelihood, the two modules are then fused within the Bayesian framework to achieve improved tracking performance in the complicated sea clutter background. Experimental results based on both the simulated and real radar data demonstrate that the proposed tracker outperforms its representative counterparts.