j. amiri parian

Detection of tree leaf diseases plays a crucial role in the horticultural field. These diseases can originate from viruses, bacteria, fungi, and other pathogens. If proper attention is not given, these diseases can drastically affect trees, reducing both the quality and quantity of yields. Due to the importance of quince in Iran's export market, its diseases can cause significant economic losses to the country. Therefore, if leaf diseases can be automatically identified, appropriate actions can be taken in advance to mitigate these losses. Traditionally, the identification and detection of tree diseases rely on experts' naked-eye observations. However, the physical condition of the expert such as eyesight, fatigue, and work pressure can affect their decision-making capability. Today, deep convolutional neural networks (DCNNs), a novel approach to image classification, have become the most crucial detection method. DCNNs improve detection or classification accuracy by developing machine-learning models with many hidden layers to extract optimal features. This approach has significantly enhanced the classification and identification of diseases affecting plants and trees. This study employs a novel CNN algorithm alongside two pre-trained models to effectively identify and classify various types of quince diseases.

Images of healthy and diseased leaves were acquired from several databases. The majority of these images were sourced from the Agricultural Research Center of Isfahan Province in Iran, supplemented by contributions from researchers who had previously studied in this field. Other supporting datasets were obtained from internet sources. This study incorporated a total of 1,600 images, which included 390 images of fire blight, 384 images of leaf blight, 406 images of powdery mildew, and 420 images of healthy leaves. Of all the images obtained, 70%, 20%, and 10% were randomly selected for the network's training, validation, and testing, respectively. Image flipping, rotation, and zooming were applied to augment the training dataset. In this research, a proposed convolutional neural network (CNN) combined with image processing was developed to classify quince leaf diseases into four distinct classes. Three CNN models, including Inception-ResNet-v2, ResNet-101, and our proposed CNN model, were investigated, and their performances were compared using essential indices including precision, sensitivity, F1-score, and accuracy. To optimize the models’ performance, the impact of dropout with a 50% probability and the number of neurons in the hidden layers were examined. Our proposed CNN model consists of an architecture with four convolutional layers, with 224 × 224 RGB images as input to the first layer, which has 16 filters, followed by additional convolutional layers with 32, 64, and 128 filters respectively. Activation functions of ReLU combined with max-pooling were used at each convolutional layer, and Softmax activation was applied in the last layer of the neural network to convert the output into a probability distribution.

Three confusion matrices based on the test dataset were constructed for all the CNN models to compare and evaluate the performance of the classifiers. The indices obtained from the confusion matrices indicated that Inception-ResNet-v2 and ResNet-101 achieved accuracies of 79% and 72%, respectively. While all models exhibited promising efficiency in classifying leaf diseases, the proposed shallow CNN model stood out with an impressive accuracy of 91%, marking it as the most effective solution. The comprehensive results indicate that the optimized CNN model, featuring four convolutional layers, one hidden layer with 64 neurons, and a dropout rate of 0.5, outperformed the transfer learning models.

The findings of this study demonstrate that our developed proposed CNN model provides a high-performance solution for the rapid identification of quince leaf diseases. It excels in real-time detection and monitoring, achieving remarkable accuracy. Notably, it can identify fire blight and powdery mildew with a precision exceeding 95%.

Control of walnut diseases and pests requires the mapping of the extent of contamination within possible shortest time. Therefore, it is necessary to develop systems to detect and determine the prevalence and location of contamination for researchers and gardeners. Image processing has been proposed as an approach to determine the extent and type of damage to different products in farms and gardens. The aim of this study was to design an algorithm based on the processing of walnut leaf images under natural light conditions in order to provide a rapid and non-destructive detection of diseases for the protection of trees using imaging methods. In this research, the possibility of detecting Anthracnose disease was investigated by processing walnut leaf images. The disease was detected using in situ images taken from the leaves to provide the basis for designing application software on smart mechatronic systems.
 

Images of leaves on walnut trees were taken under outdoor light conditions. Color and morphological properties extracted from the images were used to detect the pest on the leaves. Gnomonia leptostyla disease diagnostic algorithm was based on process of color and morphological characteristics, leaves background and disease-stained spots. The range of changes in R, G, and B indices was obtained in histograms and then two-dimensional spaces were analyzed statistically on GR, GB, and BR planes. All points from these regions were used as statistical samples, for which bivariate regressions of GR, GB, and BR were obtained as y = b0 + b1x. Segments containing anthracnose spots from the leaves were segregated by extracting the coordinates of the points in each side on the RGB color space cube. Finally, anthracnose content was detected based on the number of spots detected by the algorithms. The percentage of contamination was used to determine the amount of contamination in each imaged area.

Examination of the colored spaces indicated that the domain of the anthracnose color components on the GR side has nothing in common with the color components of the leaves. The analysis of color space data revealed that the leaves and anthracnose were more distinguishable on the GB and RB sides, respectively.According to the histogram of the HSV color space, anthracnose spots were isolated from the leaves by determining the H range. In the evaluation of the proposed method for diagnosis of anthracnose, the infection severity calculated by the algorithm with the true infection intensity. T-test results for comparing the mean of the two infection intensity samples showed no significant differences between the two groups at 1% probability level.

The evaluation of the proposed method showed a 98% segregation accuracy for G. leptostyla detection method. Based on the results, the proposed method for detecting anthracnose spots is suitable for determining the contamination severity in the imaged areas.

Repetitive and dangerous tasks such as harvesting and spraying have made robots usable in the greenhouses. The mechanical structure and navigation algorithm are two important parameters in the design and fabrication of mobile greenhouse robots. In this study, a four- wheel differential steering mobile robot was designed and constructed to act as a greenhouse robot. Then, the navigation of the robot at different levels and actual greenhouses was evaluated. The robot navigation algorithm was based on the path learning, so that the route was stored in the robot memory using a remote control based on the pulses transmitted from the wheels encoders; then, the robot automatically traversed the path. Robot navigation accuracy was tested at different surfaces (ceramics, concrete, dense soil and loose soil) in a straight path 20 meters long and a square path, 4×4 m. Then, robot navigation accuracy was investigated in a greenhouse. Robot movement deviation value was calculated using root mean square error (RMSE) and standard deviation (SD). The results showed that the RMSE of deviation of autonomous method from manual control method in the straight path to the length of 20 meters in ceramic, concrete, dense  soil and loose soil were 4.3, 2.8, 4.6 and 8 cm, and in the 4×4 m square route were 6.6, 5.5, 13.1 and 47.1 cm, respectively.