SARIVA: Smartphone APP for Real-time Intelligent Video Analytics

Abstract

This paper presents the design, implementa on and evalua on of a new smartphone applica on that is capable of real-me object detec on using both sta onary and moving cameras for embedded systems, par cularly, the Android smartphone pla orm. A new object detecon approach, Op cal ORB, is presented which is capable of real-me performance at high defini on resoluons on a smartphone. In addi on, the developed smartphone applica on has the ability to connect to a remote server and wirelessly send image frames when moving objects appear in the camera's field of view; thus, allowing the human operator to only view video frames that are of interest. Evalua on experiments show a capability of achieving real-me performance for high defini on (HD) resolu on video.

A Binary Representation for Real-Valued, Local Feature Descriptors

“…Real-valued, local feature descriptors, such as Scale-Invariant Feature Transform (SIFT) [24] and Speeded Up Robust Features (SURF) [5] have already found their place in many computer vision applications, e.g., recognition [14], localisation [11,12], tracking [8,34], simultaneous localisation and mapping [17], or retrieval [6,15]. However, there is a need for the development of more ef icient techniques, in terms of computation time, storage requirements, or robustness [6,10,12,13,25,27].…”

“…This can be seen in Binary Robust Independent Elementary Features (BRIEF) [7], where point pairs are sampled from isotropic Gaussian distribution. Binary Robust Invariant Scalable Keypoints (BRISK) [23], in turn, uses a circular pattern with equally spaced points for this purpose, and Oriented FAST and Rotated BRIEF (ORB) [8,10,33] uses a learned sampling pattern and FAST [32] technique to generate keypoints. A retinal sampling pattern is used in Fast Retina Keypoint (FREAK) [2].…”

A general purpose intelligent surveillance system for mobile devices using Deep Learning

“…In this paper, we present a general purpose surveillance system that has at its core two main elements: i) a detection module, and ii) a classification module. The detection module is based on the recently developed and published SARIVA method [2] and WhatMovesApp [3] application for detecting objects that move using smart-phone devices. The detection module was a subject of a recent publication [2] and will only briefly be outlined in section II.…”

“…The detection module is based on the recently developed and published SARIVA method [2] and WhatMovesApp [3] application for detecting objects that move using smart-phone devices. The detection module was a subject of a recent publication [2] and will only briefly be outlined in section II. The main novelty of this paper is the classification module which is based on convolutional neural network (CNN) using Deep Learning methodology [4].…”

“…Section II briefly outlines the SARIVA method used to autonomously detect any moving object. Further details on SARIVA method can be found in [2]. Section III describes the image classification module in general.…”