Deep learning to frame objects for visual target tracking

Abstract

We present a new approach to deal with visual tracking target tasks. This method uses a convolutional neural network able to rank a set of patches depending on how well the target is framed (centered). To cover the possible interferences our proposal is to feed the network with patches located in the surroundings of the object detected in the previous frame, and with different sizes, thus taking into account eventual changes of scale. In order to train the network, we had to create an ad-hoc large dataset with positive and negative examples of framed objects extracted from the Imagenet detection database. The positive examples were those containing the object in a correct frame, while the negative ones were the incorrectly framed. Finally, we select the most promising patch, using a matching function based on the deep features provided by the well-known AlexNet network. All the training stage of this method is offline, so it is fast and useful for real-time visual tracking. Experimental results show that the method is very competitive with respect to state-of-the-art algorithms, being also very robust against typical interferences during the visual target tracking process.

Hand Gesture Recognition Using Convolutional Neural Network for People Who Have Experienced A Stroke

“…CNN is defined as a multi-layer neural network with a unique architecture used for deep learning [12]. A CNN architecture is comprised of three essential layers: Convolutional Layer, Pooling Layer, and Fully-Connected Layer.…”

“…CNN is generally used in recognizing objects, scenes, and carrying out image detection, extraction and segmentation. CNN has been significantly used in the last few years ago due to the following three aspects: (1) the necessity for feature extraction by using image processing tools is removed since CNN can directly learn the image data, (2) Exceptionally good for recognition of results and can be easily re-trained for new recognition purposes, and (3) CNN can be built on the preexisting network [12].…”

Fast and Accurate Lung Tumor Spotting and Segmentation for Boundary Delineation on CT Slices in a Coarse-to-Fine Framework

“…In the past few years, deep learning has already swept through most research fields of computer vision and has achieved better results than traditional methods, and biomedical/medical image segmentation is also no exception [11,12]. The existing methods [13][14][15][16] strive to obtain more precise and comprehensive tumor features by taking the advantage of deep learning in its superior ability of hierarchical feature representation.…”

Detecting Video Surveillance Using VGG19 Convolutional Neural Networks

“…The proposed system outperforms the state-of-the-art systems. Schuchao et al [21] proposed a deep learning-based technique for tracking visual objects. They used CNN to rank the patches of the target objects based on how well it is centred.…”