Fast Visual Tracking Based on Convolutional Networks

Abstract

Recently, an upsurge of deep learning has provided a new direction for the field of computer vision and visual tracking. However, expensive offline training time and the large number of images required by deep learning have greatly hindered progress. This paper aims to further improve the computational performance of CNT which is reported to deliver 5 fps performance in visual tracking, we propose a method called Fast-CNT which differs from CNT in three aspects: firstly, an adaptive k value (rather than a constant 100) is determined for an input video; secondly, background filters used in CNT are omitted in this work to save computation time without affecting performance; thirdly, SURF feature points are used in conjunction with the particle filter to address the drift problem in CNT. Extensive experimental results on land and undersea video sequences show that Fast-CNT outperforms CNT by 2~10 times in terms of computational efficiency.

Anomalous Behaviors Detection for Underwater Fish Using AI Techniques

“…Hence coordinates and posture code of 5 +1 will be recorded in the thread of 1 . If no object matching the fish 1 was found at the next frame, then coordinates and posture code previously recorded will be used to predict [1,2] the most likely path of the target fish for (t+1) th through (t+v) th frames, where v is heuristically set to 5. For example, in Fig.…”

“…Deep learning techniques can do end-to-end detection of instances of semantic objects such as fish without specifically defining features, and are typically built on convolutional neural networks (CNN) [1,2]. Based on our previous work [3] employing deep learning Faster-rcnn [4] as an object detector to implement the tracking task for measuring the moving speed of fish, this paper presents a real-time solution for the problem of detecting anomalous behaviors for underwater fish.…”