A Comprehensive Survey of Video Datasets for Background Subtraction

Abstract

Background subtraction is an effective method of choice when it comes to detection of moving objects in videos and has been recognized as a breakthrough for the wide range of applications of intelligent video analytics (IVA). In recent years, a number of video datasets intended for background subtraction have been created to address the problem of large realistic datasets with accurate ground truth. The use of these datasets enables qualitative as well as quantitative comparisons and allows benchmarking of different algorithms. Finding the appropriate dataset is generally a cumbersome task for an exhaustive evaluation of algorithms. Therefore, we systematically survey standard video datasets and list their applicability for different applications. This paper presents a comprehensive account of public video datasets for background subtraction and attempts to cover the lack of a detailed description of each dataset. The video datasets are presented in chronological order of their appearance. Current trends of deep learning in background subtraction along with top-ranked background subtraction methods are also discussed in this paper. The survey introduced in this paper will assist researchers of the computer vision community in the selection of appropriate video dataset to evaluate their algorithms on the basis of challenging scenarios that exist in both indoor and outdoor environments.INDEX TERMS Background model, background subtraction, challenges, datasets, deep neural networks, foreground, intelligent video analytics (IVA), video frames.

Robust Vehicle Detection and Counting Algorithm Employing a Convolution Neural Network and Optical Flow

“…The proposed method for detection and counting is evaluated and compared with four state of the art algorithms [12,13,14,15]. Two experiments, including seven videos with various challenges [20], are used to validate the contribution of the proposed method. We test the proposed approach on nighttime, daytime, intermittent vehicle motion, and crowd scenes, as mentioned in Table 3.…”

HSMD: An Object Motion Detection Algorithm Using a Hybrid Spiking Neural Network Architecture

“…Motion detection is performed by analysing/tracking the variation of light intensities between a set of image frames. Camera, background and foreground are three factors that affect the quality of the BS [18]. Current BS challenges include (i) abrupt illumination changes, which impact the pixel intensity values and may increase the number of false positives; (ii) dynamic objects, where background object movement may interfere with motion detection of static BS; (iii) relative motion, where both the camera and the object move at the same time, creating dynamic backgrounds; (iv) challenging weather conditions such as fog, rain, snow, air or turbulence generates errors; (v) camouflage, where camouflage regions occur when the foreground and background light intensity pixels are similar; (vi) occlusion, when another object or fixed structure obstructs the object of interest; (vii) irregular object motion -objects that suddenly increase or decrease in speed; (viii) noise, possibly arising from dirty lenses, dust, extremely high/low light intensity, etc.…”

Comparative Monte Carlo Analysis of Background Estimation Algorithms for Unmanned Aerial Vehicle Detection

“…Sometimes, evaluation of the algorithms uses additional criteria, such as CPU computing requirements or the amount of memory needed [33]. There are also papers describing the databases used to evaluate background estimation algorithms [35]. Metaanalyses related to the use of background estimation algorithms constitute an interesting source of knowledge related to the approaches of various researchers [36].…”