Immature peach detection in colour images acquired in natural illumination conditions using statistical classifiers and neural network

Sensors and systems for fruit detection and localization: A review

“…Slaughter and Harrell (1989) used color camera to acquire images and detect oranges based on 'HS' components of Hue, Saturation, Intensity (HSI) color space and Red, Green, Blue channels of RGB color spaces of objects and achieved an classification accuracy of 75%. Similarly, Kurtulmus et al (2014) used color camera for detection of peaches using statistical classifier and neural network. The major disadvantage of this sensor is that the images captured are sensitive to variation in lighting conditions.…”

“…Majority voting of a K number of nearest neighbor are used for classification (Shapiro and Stockman, 2001). Various researchers including Linker et al (2012), Seng and Mirisaee (2009) and Kurtulmus et al (2014) have used KNN clustering classification to classify different fruit. Linker et al (2012) used KNN classifier for classification of green apples.…”

Improvements to and large‐scale evaluation of a robotic kiwifruit harvester

“…A wide range of sensor types have been investigated for a variety of fruit types in an attempt to effectively locate fruit and flowers in these challenging conditions (Gongal, Amatya, Karkee, Zhang, & Lewis, ). The most common approach to fruit detection utilises colour cameras as they provide a range of information for detection, including colour, geometric and textural information about the fruit (Kurtulmus, Lee, & Vardar, ). The dynamic lighting conditions make traditional image computing algorithms unreliable (Gongal et al, ); however, soft computing approaches have shown strong results.…”

Detecting tomatoes in greenhouse scenes by combining AdaBoost classifier and colour analysis

“…In this work, a SVM which uses a linear kernel and a BP neural network which were constructed using the training set containing two classes. Table 1 also lists the true positive rates, false negative rates and false positive rates of the three referencing approaches (Kurtulmus et al, 2014;Zhao et al, 2016). It can be seen that the proposed tomato detection algorithm achieved b i o s y s t e m s e n g i n e e r i n g 1 4 8 ( 2 0 1 6 ) 1 2 7 e1 3 7 the second highest true positive rate while it maintained the second lowest false positive rate which is 0.1%, slightly lower than that of feature images fusion approach.…”

“…About 80% of fruit pixels were correctly classified under all lighting conditions with a less than 3% error rate. Kurtulmus, Lee, and Vardar (2014) also proposed an algorithm based on support vector machine (SVM) to detect cutting locations of corn tassels in natural outdoor maize canopy. The author argued that the proposed algorithm performed with a correct detection rate of 81.6% for the test set.…”