Intelligent detection of imperfect wheat grains based on machine vision is of great significance to correctly and rapidly evaluate wheat quality. There is little difference between the partial characteristics of imperfect and perfect wheat grains, which is a key factor limiting the classification and recognition accuracy of imperfect wheat based on a deep learning network model. In this paper, we propose a method for imperfect wheat grains recognition combined with an attention mechanism and residual network (ResNet), and verify its recognition accuracy by adding an attention mechanism module into different depths of residual network. Five residual networks with different depths (18, 34, 50, 101, and 152) were selected for the experiment, it was found that the recognition accuracy of each network model was improved with the attention mechanism, and the average recognition rate of ResNet-50 with the addition of the attention mechanism reached 96.5%. For ResNet-50 with the attention mechanism, the optimal learning rate was further screened as 0.0003. The average recognition accuracy reached 97.5%, among which the recognition rates of scab wheat grains, insect-damaged wheat grains, sprouted wheat grains, mildew wheat grains, broken wheat grains, and perfect wheat grains reached 97%, 99%, 99%, 95%, 96%, and 99% respectively. This work can provide guidance for the detection and recognition of imperfect wheat grains using machine vision.

For the efficient selection of high-quality rice varieties, the near-infrared spectroscopy (NIRS) technique has been widely applied to detect constituents in single rice kernels. Compared with other constituents, amylose content (AC) and fat content (FC) are the key parameters that can affect the quality of rice. Based on two modified AC and FC trace detection methods, two NIRS methods to detect AC and FC in single rice kernels were developed. Using the proposed methods, the AC and FC in two groups of rice kernel datasets were measured. The datasets were collected on two spectrometers with different sample movement states (static and dynamic) and measurement modes (diffuse reflectance (NIRr) and diffuse transmission (NIRt)). By optimizing the pre-treatment method and spectral range, the determination coefficients of cross-validation (R²_cv) and prediction (R²_p) of the NIRS models under different measurement conditions were all above 0.6. The results indicated that the proposed methods were applicable to the rapid, non-destructive detection and sorting of individual rice seeds with different AC and FC, and it was shown that these methods can meet the requirements of the rough screening of rice seed varieties.