Abstract

Abstract
The choroid is the middle layer of the eye globe located between the retina and the sclera. It is proven that choroidal thickness is a sign of multiple eye diseases. Optical Coherence Tomography (OCT) is an imaging technique that allows the visualization of tomographic images of near surface tissues like those in the eye globe. The automatic calculation of the choroidal thickness reduces the subjectivity of manual image analysis as well as the time of large scale measurements. In this paper, a method for the automatic estimation of the choroidal thickness from OCT images is presented. The pre-processing of the images is focused on noise reduction, shadow removal and contrast adjustment. The inner and outer boundaries of the choroid are delineated sequentially, resorting to a minimum path algorithm supported by new dedicated cost matrices. The choroidal thickness is given by the distance between the two boundaries. The data are then interpolated and mapped to an infrared image of the eye fundus. The method was evaluated by calculating the error as the distance from the automatically estimated boundaries to the boundaries delineated by an ophthalmologist. The error of the automatic segmentation was low and comparable to the differences between manual segmentations from different ophthalmologists.

Abstract
In this paper, we introduce the streaming graph canonization problem. Its goal is finding a canonical representation of a sequence of graphs in a stream. Our model of a stream fixes the graph's vertices and allows for fully dynamic edge changes, meaning it permits both addition and removal of edges. Our focus is on small graphs, since small graph isomorphism is an important primitive of many subgraph-based metrics, like motif analysis or frequent subgraph mining. We present an efficient data structure to approach this problem, namely a graph isomorphism discrete finite automaton and showcase its efficiency when compared to a non-streaming-aware method that simply recomputes the isomorphism information from scratch in each iteration.

Abstract
Many cloud vision APIs are available on the internet to recognize emotion from facial images and video analysis. The capacity to recognize emotions under various poses is a fundamental requirement in the area of emotion recognition. In this paper, the performance of two famous emotion recognition APIs is evaluated under the facial images of various poses. The experiments were done with the public dataset containing 980 images of each type of five poses [full left, half-left, straight, half-right, and full-right] with the seven emotions (Anger, Afraid, Disgust, Happiness, Neutral, Sadness, Surprise). It has been discovered that overall recognition accuracy is best in Microsoft Azure for straight images, whereas the face detection capability is better in Google. The Microsoft did not detect almost any of the images with full left and full right profile, but Google detected almost all of them. The Microsoft API presents an average true positive value up to 60%, whereas Google presents the maximum true positive value 45.25%.

Abstract
This paper presents a Pattern Recognition System, which can be used in classification applications for hand gestures for control of robotic arms. The system based in three steps, uses feature matching for extracting objects from a scene, edge detector and deep learning. The use of extraction of the region of interest and edges segmentation reduces the amount of processing required to recognize signals, thus speeding up the recognition process. Experimental classification results were positive with good statistical results. The presented data were tested considering four different types of segmentation implementations.

Abstract