Hyperspectral imaging (HSI) is composed of both feature noise and label noise; thus, makes HSI crop classification extremely challenging. Recently, various feature noise tolerant machine learning and Deep learning mechanism have been presented for HSI crop classification considering; however, very limited have been presented considering presence of noise in training label. Existing model applied with presence of both label and feature noise in HSI exhibit very poor accuracies with higher misclassification. In addressing the research challenges in this paper noisy and mixedpixel aware (NMA) HSI crop classification is proposed. The NMA-HSI provide a mechanism to obtain good quality spatial-spectral feature under presence of feature and label noise. The work introduces a modified SVM hyperplane to address class imbalance issues considering presence of noisy and mixed pixel within HSI. Experiment outcome shows the NMA-HSI achieves higher overall accuracies, average accuracies, and Kappa in comparison with existing HSI crop classification model using standard HSI datasets.