Automatic defect classification methods are important to increase the productivity of the forest industry. In this respect, classification is also an important component of a pattern recognition system. Well designated classification algorithm will make recognition process more efficient and productive. Quality control is one of the most important steps among the applications that use classification. There are various techniques which are available in order to check quality of wooden material. However, display based quality control of wooden materials is still stands as a hard process. Although there are quality control methods, there are quality control problems because of the classification depending on the person's eyestrain. We aim the reduction of labor costs. Also, we aim to achieve a more accurate and reliable way to make automatic quality classification. In this article, we proposed a method for quality control of wood material. In order to determine quality control of wooden material, knot detection algorithm which is developed using image processing techniques. This knot detection algorithm consists of several steps. These steps are morphological preprocesses in the knot preprocessing step, knot features obtained from Wavelet Moment (WM) in the feature extraction step, k nearest neighbor method (KNN) classification technique in the classification step. Index Terms-Approach coefficients; knot defect types; k nearest neighbour method; wavelet moment; wood.

In the study carried out in line with the stated purposes, monthly rain, humidity and temperature data, wheat production amount, and wheat productivity data of Konya province between 1980-2020 were used. Using these data, wheat productivity estimation was performed with (Gated Recurrent Units) GRU and Long Short Term Memory (LSTM) methods, which are Recurrent Neural Network (RNN) based algorithms. When wheat productivity estimation performance was examined with the implemented GRU-based model, 0.9550, 0.0059, 0.0280, 0.0623, 7.45 values were obtained for the R 2 score, MSE, RMSE, MAE and MAPE values, respectively. In the performance results obtained with the LSTM method, which is another RNN-based method, 0.9667, 0.0054, 0.0280, 0.0614, 7.33 values were obtained for the R 2 score, MSE, RMSE, MAE and MAPE values, respectively. Although the LSTM method gave better results than the GRU method, the training modelling time of the LSTM method took longer than that of the GRU method.

No abstract

It is difficult to constantly control apple trees in farmland. In case of a disease on tree leaves, the risk of disease transmission to other leaves is high. It is necessary to prevent further deterioration of the plant by performing automatic detection of the disease in the early period. If the disease detection is delayed, the planned production cannot be realized. It is too late if diseases are detected by a farmer or agronomist. In addition, as the agricultural lands grow, the number of experts needed increases accordingly. For these reasons, leaf images of apple trees are grouped into 4 different classes: apple peel, leaf rust, healthy apple and multiple disease states. In the proposed method, noise removal in the images, detection of the relevant area and histogram equalization on the YUV color space are performed. Due to the unbalanced class distribution in the data set used, data augmentation was applied for the minority classes with the SMOTE method. Afterwards, features are extracted using pre-trained network models DenseNet121, DenseNet201, InceptionResNetV2, InceptionV3, ResNet50V2. Extracted features were classified with a CNN-based method developed with an accuracy of 99%.

No abstract