One of the major issues in handwritten character recognition is the efficient creation of ground truth to train and test the different recognizers. The manual labeling of the data by a human expert is a tedious and costly procedure. In this paper we propose an efficient and low-cost semiautomatic labeling system for character datasets. First, the data is represented in different abstraction levels, which is clustered after in an unsupervised manner. The different clusters are labeled by the human experts and finally an unanimity voting is considered to decide if a label is accepted or not. The experimental results prove that labeling only less than 0.5% of the training data is sufficient to achieve 86.21% recognition rate for a brand new script (Lampung) and 94.81% for the MNIST benchmark dataset, considering only a K-nearest neighbor classifier for recognition.

No abstract

Application Programming Interface (API) is an interface built by system developer so some or entire functions of the system can programatically be accessed. Representational State Transfer (REST) is one of API development architectural style that uses Hypertext Transfer Protocol (HTTP) for data communication. This research implemented REST in developing API as the back-end of the skincare clinic patient information system. API was developed using Javascript Object Notation (JSON) as the standard format for data communication and JSON Web Token (JWT) as user authentication code. This research indicates that the development of API successfully performed on the patient administration of skin care clinic and implementation of REST makes it easy to develop API structures. This research produced REST API-based back-end for the patient administration information system of skin care clinic. API was tested in three stages: JWT testing on multiple backend servers, API testing with Equivalence Partitioning and system functional testing.

Gita Persada Butterfly Park is the only breeding of engineered in situ butterflies in Indonesia. It is located in Lampung and has approximately 211 species of breeding butterflies. Each species of Butterflies has a different texture on its wings. The Limited ability of the human eye to distinguishing typical textures on butterfly species is the reason for conducting a research on butterfly identification based on pattern recognition. The dataset consists of 600 images of butterfly’s upper wing from six species: Centhosia penthesilea, Papilio memnon, Papilio nephelus, Pachliopta aristolochiae, Papilio peranthus and Troides helena. The pre-processing stage is conducted using scaling, segmentation and grayscale methods. The GLCM method is used to recognize the characteristics of butterfly images using pixel distance  and Angular direction 0o, 45o, 90o and 135o. The features used is angular second moment, contrast, homogeneity and correlation. KNN classification method in this study uses k values1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21 and 23 based on the Rule of Thumb. The result of this study indicate that Centhosia penthesilea and Papilio nephelus classes can be classified properly compared to the other 4 classes and require a classification time of 2 seconds at each angular orientation. The highest accuracy is 91.1% with a value of  in the angle of 90o and error rate8.9%. Classification error occured because the value of the test data features is more dominant with the value of the training image features in different classes than the supposed class.  Another reason is because of imperfect test data.

Batik is a famous name of a traditional fabric from Java. It has been admitted as one if the traditional cultural heritage of Indonesia by UNESCO since October 2nd, 2009. Over the time, Batik is copied and modified by many regions in Indonesia resulting some new unique motifs. Batik Lampung is an sample of them. This paper deals with the k-Nearest Neighbor classification of the motifs (pattern) of the Batik Lampung. The known motifs of Batik Lampung consist of Jung Agung, Siger Kembang Cengkih, Siger Ratu Agung, and Sembagi. The original image samples are stored in RGB. They are firstly resized into 50 x 50 pixels and then converted to grayscale image. To recognize them, the Gray Level Co-Occurence Matrix (GLCM) feature is extracted and k-Nearest Neighbor (k-NN) with values of k = 3, 5, 7, 9, 11 and orientation angle of 00 450, 900, 1350 is applied to classify the motifs. The best accuracy is achieved at the rate 97,96% for k = 7 and angle1350.