Multi-Stage Corn Yield Prediction Using High-Resolution UAV Multispectral Data and Machine Learning Models
Round 1
Reviewer 1 Report
There is no clear comparison between the five different ML algorithms including Linear Regression (LR), k-Nearest Neighbor (KNN), Random Forest (RF), Support Vector Regression (SVR), and Artificial Neural Network (DNN) were used for yield prediction.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
1. In line 85 of the abstract, the given examples do not show that the training data, input variables, crop type, and growth stage play a decisive role in the performance of the ML model.
2. Line 197, briefly explaining why these five were chosen as experimental methods.
3. Does the gripper camera recognize all the objects identified by the top camera again? If so, does the gripper plan its path or need to return to its initial position each time it reaches the target object?
4. Line 415, Figure 7 Figure 8 shows not the impact of the selected variables on the model performance, but the impact of the number of selected variables on the model performance.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Multi-Stage Corn Yield Prediction Using High-Resolution UAV Multispectral Data and Machine Learning Models
This research aimed to evaluate the effectiveness of the UAV multispectral data and various machine learning (ML) models for for predicting corn yield at the farm level with a limited number of training samples. The results show that the spectral bands and vegetation indices indicated moderate to good correlation with the yield at the vegetative (V6) and reproductive (V5) growth stage indicating their suitability to be used in predicting yield using ML models. Furthermore, the comparative analysis of ML models demonstrated that Support Vector Regression (SVR) and k-Nearest Neighbor (KNN) outperformed other models at both growth stages.
Although the topic of the manuscript is of wide interest in the scientific community and for this journal, I am seeing several minor methodological and scientific writing flaws in this study.
Review summary
· Please emphasize the main objectives of the research in the last paragraph, e.g., from LN 106 – 115. At the end of the Introduction, the main study objectives must be clearly indicated
· Figure 2 indicates hyperparameter optimization, however, the Readers do not get the information about the final optimized values, e.g., I want to know the mtry parameter of the RF model, since only 30 different values were used..
· In my opinion, the results for the DNN method are already known before the study was conducted since neural networks need a lot of input training data, and this research, as indicated by the Authors, used limited number of training samples
· Please adjust throughout the manuscript terms Artificial Neural Networks and Deep Neural Network
At the same time, the attempted methodology is very actual and interesting, but the Authors have to emphasize the main objectives at the beginning of the manuscript. My final opinion is that this research has a solid potential and it is very interesting, but needs some additional corrections.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf