Ferhat Kurtulmuş

Detecting corn tassels using computer vision and support vector machines

An automated solution for maize detasseling is very important for maize growers who want to reduce production costs. Quality assurance of maize requires constantly monitoring production fields to ensure that only hybrid seed is produced. To achieve this cross-pollination, tassels of female plants have to be removed for ensuring all the pollen for producing the seed crop comes from the male rows. This removal process is called detasseling. Computer vision methods could help positioning the cutting locations of tassels to achieve a more precise detasseling process in a row. In this study, a computer vision algorithm was developed to detect cutting locations of corn tassels in natural outdoor maize canopy using conventional color images and computer vision with a minimum number of false positives. Proposed algorithm used color informations with a support vector classifier for image binarization. A number of morphological operations were implemented to determine potential tassel locations. Shape and texture features were used to reduce false positives. A hierarchical clustering method was utilized to merge multiple detections for the same tassel and to determine the final locations of tassels. Proposed algorithm performed with a correct detection rate of 81.6% for the test set. Detection of maize tassels in natural canopy images is a quite difficult task due to various backgrounds, different illuminations, occlusions, shadowed regions, and color similarities. The results of the study indicated that detecting cut location of corn tassels is feasible using regular color images.

Discriminating rapeseed varieties using computer vision and machine learning

We developed an expert system for detection of rapeseed variety using computer vision and machine learning.Scanner images can be used to discriminate rapeseed variety.Developed algorithm determines successfully variety of rapeseed with an accuracy rate of 99.24%. Rapeseed is widely cultivated throughout the world for the production of animal feed, vegetable fat for human consumption, and biodiesel. Since the seeds are evaluated in many areas for sowing and oilseed processing, they must be identified quickly and accurately for selection of a correct variety. An affordable method based on computer vision and machine learning was proposed to classify the seven rapeseed varieties. Different types of feature sets, feature models, and machine learning classifiers were investigated to obtain the best predictive model for rapeseed classification. The training and test sets were used to tune the model parameters during the training epochs by varying the complexity of the predictive models with grid-search and K-fold cross validation. After obtaining optimized models for each level of complexity, a dedicated validation set was used to validate predictive models. The developed computer vision system provided an overall accuracy rate of 99.24% for the best predictive model in discriminating rapeseed variety.

Classification of chestnuts according to moisture levels using impact sound analysis and machine learning

In this study, a prototype system was designed, built and tested to classify chestnuts using impact sound signals and machine learning methods according to moisture contents. Briefly, the system consisted of a shotgun microphone, a sliding platform, an impact surface, a triggering system, a sound device and a computer. Sound signal data were acquired from 2028 chestnut samples with three different moisture levels. Acoustic signals from chestnut samples were filtered to alleviate negative effects of unwanted noise. Four machine learning classifiers using three different feature sets obtained from two feature groups applying feature reduction methods were trained and tested to classify pairs of chestnut moisture group categories as 35% versus 45%, 35% versus 55%, 45% versus 55% (classification with two outputs) and 35% versus 45% versus 55% (classification with three outputs), respectively. The highest classification success (88%) was achieved for the classification application category of 35 versus 55%.