Sketch recognition

About: Sketch recognition is a research topic. Over the lifetime, 1611 publications have been published within this topic receiving 40284 citations.

A Gesture Recognition System for Cranes Using Deep Learning with a Self-attention Mechanism

Abstract: This research is aimed at recognizing the gesture of a lifting coordinator and automating the operation of a crane by introducing a system with deep learning. This paper first explains the outline of a gesture recognition system, and describes skeletal detection and its accuracy improvement technique. Furthermore, a gesture recognition system is constructed using a 1DCNN, and the recognition accuracy is verified to be improved by introducing a self-attention mechanism.

Geometric modeling of non-rigid 3D shapes : theory and application to object recognition.

Abstract: GEOMETRIC MODELING OF NON-RIGID 3D SHAPES: THEORY AND APPLICATION TO OBJECT RECOGNITION Mostafa Aly Ahmed Abdelrahman

A Sketch Recognition Algorithm Based on Bayesian Network and Convolution Neural Network

Abstract: Most of the existing sketch recognition algorithms are used to restrict the user’s drawing habits to achieve the stroke grouping and recognition. In order to solve the problem, a new sketch recognition algorithm based on Bayesian network and convolution neural network (CNN) is proposed. First, the input sketch is processed by Gaussian low-pass filter and a smoother stroke can be obtained. The stroke of continuous input is divided, then the Bayesian network and CNN are performed on stroke recognition respectively. The recognition result of Bayesian network is adopted when the reliability of stroke is larger than the threshold, otherwise recognition result of CNN will be adopted. The experiment result shows that the proposed algorithm is effective in circuit symbol recognition. The recognition rate was achieved 80.34% in the drawing process, and the final recognition rate was achieved 93.48%.

The importance of context and semantic descriptions in object recognition: Studies in computer vision and human vision

Abstract: Object recognition and scene classification are among the main interests in computer vision which have been investigated for long. Automatic recognition and classification of objects and scenes is an important skill to be gained by computers, especially in the field of artificial intelligence. Merging this skill with the ever increasing computing power of the computers will help in the development of many applications that are yet to be resolved. In this article, we present a survey on contextual and semantic approaches for object recognition by reviewing both computer vision and human vision literatures.

Penetration points method

Abstract: Finding a method which allows a computer recognition to be close to human recognition is a goal of many works in the present. We have set this goal too. According to us, we need to find function for simple recognition of shapes in the images as first step of this goal. Result of this method provides input of our system of recognition. System form depends on the result of shape recognition method. In the present, we only know that this system is going to be based on grammars. Probably it will be sub-grammar of shape grammar (now, this is only ours postulate). In this paper we present our experiment with method of simple shape recognition. This experiment is divided to several parts from all are described in this paper. Overall, we can describe these parts as a sequence: selection of shape recognition method, implementation of this method and verification of this method on the set of tests.