Object counting method based on dual attention network

Visual Attention Models of Object Counting

We develop a sequential learning model using a recurrent neural network architecture and reinforcement learning to recognize and count objects in images. Simple feedforward neural networks perform well on this task when trained using backpropagation; however, convolutional neural networks are computationally expensive and results are less certain when the image input has imperfect resolution ou...

An Object Extraction Method Based on Oscillatory Network

Based on locally excitatory globally inhibitory oscillator networks (LEGION), an object extraction method was proposed. The oscillation network can segment image and extracted objects by the scene and object feature. In this paper, the normalized difference water index (NDWI) was used as the feature of water body to segment remote sensing image. Because of the good coherence of spectrum propert...

A Visual Attention Based Method for Object Tracking

Object tracking is a very important operation in many surveillance applications. It is also closely related to motion detection/estimation and object recognition. This paper proposes a visual attention based method for object tracking. This technique is used to generate motion vectors for each frame in a moving video sequence. Results are compared with MPEG video encoder produced motion vectors...

Attention-based Object Detection

Dual Attention Network for Visual Question Answering

Visual Question Answering (VQA) is a popular research problem that involves inferring answers to natural language questions about a given visual scene. Recent neural network approaches to VQA use attention to select relevant image features based on the question. In this paper, we propose a novel Dual Attention Network (DAN) that not only attends to image features, but also to question features....