In order to compensate for soils that are deficient in certain minerals, contemporary fertilisation techniques look for signs of nutritional deficiencies in plant leaves. In order to create a fertilisation strategy that supplies the vital micronutrients without overproducing others, it is important to determine the amount of vitamins and minerals that will be required. In agriculture, nutritional deficiencies are indicated by plant leaves. It is possible to physically examine these leaves from the plant's morphological components, which gives enough information to identify the nutritional shortage. The primary objective of this study was to identify nutritional deficiencies in grape plant leaves. In this research, we use Convolutional Neural Networks (CNNs) for model training. By feeding the model's output into an LSTM, we can classify the images of grape plant leaves as either healthy, N-deficient, Fe-Deficiency, P-Deficiency, Mn-Deficiency, K-Deficiency, Zn-Deficiency, Ca-Deficiency, B-Deficiency, Mg-Deficiency, or S-Deficiency. Accuracy measures the model's performance. Data analysis revealed that the suggested model had an accuracy rate of 98.6 percent. These results suggest that convolutional neural networks (CNNs) might be useful for identifying nutritional deficits in grape plant leaves.