Various industrial plants are being cultivated in the world to produce food or industrial materials. Among these, sugarcane is one of the plants that are used and cultivated both for the extraction of sugar materials and also for the paper-making, alcohol-making, and chemical industries. This product, like other plants, can be affected by various diseases and pests. Any damage in this product may reduce the yield of produced sugar and its byproducts. Since sugarcane is an industrial cultivation, the pests and diseases of this product are also revealed on a wide scale and sometimes it is very difficult to control and manage them. Therefore, it can be mention that if the involved farmers in sugarcane cultivation units are familiar with the pests and diseases of this product, they can identify them in time and start management measures against them.
Following the destruction of a major part of production due to diseases, one of the common challenges around the world is to focus on effective methods of disease detection in plants. Hereupon, there is a need for early identification of plant disease. In recent years, widespread studies have been done to identify plant diseases and they are still ongoing.
Various pathogens such as Rhizoctonia, Cytospora and Glomerella species can settle on the pods and stalks of sugarcane and cause infection. The initial symptoms of these diseases are usually observed in the form of burnt spots on the pods or stalks. In some cases, it is possible that the disease causes dry rot on the pods. Some of these pathogens may turn the infected area brown, orange or red, and sometimes these rots are known as red rot of sugarcane. Red rot was first reported as a disease of sugarcane in Java in 1893 (Went et al. 1893). Red rot is one of the most serious diseases of sugarcane in many countries including India, Pakistan, Bangladesh, Thailand, Myanmar, Nepal, Vietnam and other countries.
Automatic plant disease classification algorithms are very important in agriculture. Early and fast identification of plant diseases is difficult due to the lack of necessary infrastructure in many parts of the world. Using modern technologies, high-quality images of smartphone are available worldwide. Ghasemi et al. (2017) used a system based on image processing method and hybrid neural network model in order to diagnose three diseases of leaves in apple (black spot, Alternaria and leafminer). Indeed, image processing was used to process and extract the characteristics of each of the images; and the hybrid artificial neural network was used to classify the diseases. Two particle swarm optimization algorithms (PSO) and Levenberg–Marquardt algorithm (LM) were used to train the model. It was observed that this system has a good performance in diagnosing the aforementioned disease with 33% accuracy and R=0.381 and RMSE=0.033 indicators. Masoumi et al. (2006) studied on mosaic virus of sugarcane using molecular analysis and identified them as A-SCMV strain. Hemalatha et al. (2020) conducted studies on sugarcane diseases using deep learning and they were able to identify the diseases including red rust, yellow spots, Helmantospora leaf spot, Cercospora leaf spot and red rot. Around 3000 images were trained. The accuracy of 96% was obtained. Thilagavathi et al. (2020) recognized different diseases in sugarcane leaves using image processing. They used Adaptive histogram equalization (AHE) for segmentation and statistical features such as variance, skewness, standard deviation, mean and covariance were extracted using gray level covariance matrix (GLCM) and principal component analysis (PCA). Then, the classifier was assessed by support vector machine (SVM). At the end, average accuracy of 95% was obtained.
In Iran, sugarcane cultivation has reached 7 million tons with a growth of 70% in 2016 (Nikpay and François-Régis, 2016). Sugarcane cultivation is expanding in different regions of Iran which is being cultivated mainly in Khuzestan province. Many agricultural and industrial complexes have been established for the cultivation and processing of sugarcane. The importance of sugarcane is so high in economy and lots of job opportunities have been created. Therefore, it is necessary to know and control pathogenic agents. The aim of this research is to identify one of the famous diseases named red rot using three convolutional neural networks namely ResNet, DenseNet and VGG.

First, 1200 images of sugarcane were collected. 70%, 20% and 10% of the data was allocated to train set, validation and test set, respectively. Figure 1 shows a number of samples under training. Also, see figure 2 for classes of healthy and infected by red rot.
Various industrial plants are being cultivated in the world to produce food or industrial materials. Among these, sugarcane is one of the plants that are used and cultivated both for the extraction of sugar materials and also for the paper-making, alcohol-making, and chemical industries. This product, like other plants, can be affected by various diseases and pests. Any damage in this product may reduce the yield of produced sugar and its byproducts. Since sugarcane is an industrial cultivation, the pests and diseases of this product are also revealed on a wide scale and sometimes it is very difficult to control and manage them. Therefore, it can be mention that if the involved farmers in sugarcane cultivation units are familiar with the pests and diseases of this product, they can identify them in time and start management measures against them.
Following the destruction of a major part of production due to diseases, one of the common challenges around the world is to focus on effective methods of disease detection in plants. Hereupon, there is a need for early identification of plant disease. In recent years, widespread studies have been done to identify plant diseases and they are still ongoing.
Various pathogens such as Rhizoctonia, Cytospora and Glomerella species can settle on the pods and stalks of sugarcane and cause infection. The initial symptoms of these diseases are usually observed in the form of burnt spots on the pods or stalks. In some cases, it is possible that the disease causes dry rot on the pods. Some of these pathogens may turn the infected area brown, orange or red, and sometimes these rots are known as red rot of sugarcane. Red rot was first reported as a disease of sugarcane in Java in 1893 (Went et al. 1893). Red rot is one of the most serious diseases of sugarcane in many countries including India, Pakistan, Bangladesh, Thailand, Myanmar, Nepal, Vietnam and other countries.
Automatic plant disease classification algorithms are very important in agriculture. Early and fast identification of plant diseases is difficult due to the lack of necessary infrastructure in many parts of the world. Using modern technologies, high-quality images of smartphone are available worldwide. Ghasemi et al. (2017) used a system based on image processing method and hybrid neural network model in order to diagnose three diseases of leaves in apple (black spot, Alternaria and leafminer). Indeed, image processing was used to process and extract the characteristics of each of the images; and the hybrid artificial neural network was used to classify the diseases. Two particle swarm optimization algorithms (PSO) and Levenberg–Marquardt algorithm (LM) were used to train the model. It was observed that this system has a good performance in diagnosing the aforementioned disease with 33% accuracy and R=0.381 and RMSE=0.033 indicators. Masoumi et al. (2006) studied on mosaic virus of sugarcane using molecular analysis and identified them as A-SCMV strain. Hemalatha et al. (2020) conducted studies on sugarcane diseases using deep learning and they were able to identify the diseases including red rust, yellow spots, Helmantospora leaf spot, Cercospora leaf spot and red rot. Around 3000 images were trained. The accuracy of 96% was obtained. Thilagavathi et al. (2020) recognized different diseases in sugarcane leaves using image processing. They used Adaptive histogram equalization (AHE) for segmentation and statistical features such as variance, skewness, standard deviation, mean and covariance were extracted using gray level covariance matrix (GLCM) and principal component analysis (PCA). Then, the classifier was assessed by support vector machine (SVM). At the end, average accuracy of 95% was obtained.
In Iran, sugarcane cultivation has reached 7 million tons with a growth of 70% in 2016 (Nikpay and François-Régis, 2016). Sugarcane cultivation is expanding in different regions of Iran which is being cultivated mainly in Khuzestan province. Many agricultural and industrial complexes have been established for the cultivation and processing of sugarcane. The importance of sugarcane is so high in economy and lots of job opportunities have been created. Therefore, it is necessary to know and control pathogenic agents. The aim of this research is to identify one of the famous diseases named red rot using three convolutional neural networks namely ResNet, DenseNet and VGG.

First, 1200 images of sugarcane were collected. 70%, 20% and 10% of the data was allocated to train set, validation and test set, respectively. Figure 1 shows a number of samples under training. Also, see figure 2 for classes of healthy and infected by red rot.