Crop disease detection and crop baking stage judgement require large image data to improve accuracy. However, the existing crop disease image datasets have high asymmetry, and the poor baking environment leads to image acquisition difficulties and colour distortion. Therefore, we explore the potential of the self-attention mechanism on crop image datasets and propose an innovative crop image data-enhancement method for recurrent generative adversarial networks (GANs) fused with the self-attention mechanism to significantly enhance the perception and information capture capabilities of recurrent GANs. By introducing the self-attention mechanism module, the cycle-consistent GAN (CycleGAN) is more adept at capturing the internal correlations and dependencies of image data, thus more effectively capturing the critical information among image data. Furthermore, we propose a new enhanced loss function for crop image data to optimise the model performance and meet specific task requirements. We further investigate crop image data enhancement in different contexts to validate the performance and stability of the model. The experimental results show that, the peak signal-to-noise ratio of the SM-CycleGAN for tobacco images and tea leaf disease images are improved by 2.13% and 3.55%, and the structural similarity index measure is improved by 1.16% and 2.48% compared to CycleGAN, respectively.

Accurate identification of the continuous physiological states of crops and diseases and pests is crucial for agricultural production management. However, real agricultural monitoring scenarios often encounter problems such as high similarity in visual differences between classes, complex background interference, and unstable lighting and environmental conditions. Additionally, due to the limited computing power, storage space, and power consumption budget of edge devices, traditional methods with high parameter volume and high computational complexity are difficult to meet the requirements of real-time monitoring and end-device deployment. To address these issues, we propose an efficient and lightweight deep perception method for complex agricultural visual tasks, Frequency-Spatial Enhanced Lightweight ConvNeXt (FSEL-NeXt), which mainly consists of the Vivid Enhanced Spatial Attention Stem (VESAS) and the Frequency-Spatial Enhancement and Dilation (FSED). VESAS retains high-resolution texture information and extracts multi-scale semantic features through a dual-branch structure; FSED combines asymmetric spatial convolution with frequency-domain multi-layer perceptron to jointly model local texture, global morphology, and frequency information, thereby improving the model's ability to recognize subtle category differences with lower computational cost. We conducted comprehensive evaluations on multiple datasets including tobacco drying status, banana ripeness, and plant diseases.