Histogram of Fuzzy Local Spatio-Temporal Descriptors for Video Action Recognition

Abstract

Feature extraction plays a vital role in visual action recognition. Many existing gradient-based feature extractors, including histogram of oriented gradients (HOG), histogram of optical flow (HOF), motion boundary histograms (MBH), and histogram of motion gradients (HMG), build histograms for representing different actions over the spatio-temporal domain in a video. However, these methods require to set the number of bins for information aggregation in advance. Varying numbers of bins usually lead to inherent uncertainty within the process of pixel voting with regard to the bins in the histogram. This paper proposes a novel method to handle such uncertainty by fuzzifying these feature extractors. The proposed approach has two advantages: i) it better represents the ambiguous boundaries between the bins and thus the fuzziness of the spatio-temporal visual information entailed in videos, and ii) the contribution of each pixel is flexibly controlled by a fuzziness parameter for various scenarios. The proposed family of fuzzy descriptors and a combination of them were evaluated on two publicly available datasets, demonstrating that the proposed approach outperforms the original counterparts and other state-of-the-art methods.

A distributed Content-Based Video Retrieval system for large datasets

“…According to Table 2, the proposed approach has achieved higher accuracy for this dataset when compared to Yan et al [76] and Banerjee et al [77] which achieves 41.54% and 78% accuracy, respectively, but it does not exceed the results presented in Zuo et al [78]. Figure 14 shows that most of the classes achieved more than 80% accuracy results, which explains the efficiency of our approach in the real-world activity recognition process.…”

“…Using the UCF50 dataset, the proposed method is compared with three similar approaches, including Yan et al [76], Banerjee et al [77] and Zuo et al [78] . For all dataset classes, the videos are split into 25 groups.…”

Seen to Unseen: When Fuzzy Inference System Predicts IoT Device Positioning Labels That Had Not Appeared in Training Phase

“…In contrast, ML and Deep Learning (DL) [12] were employed to transfer an indoor tracking problem into a classification problem. In recent year, ML and DL have achieved great success with a wide spectrum of applications including action recognition in videos [13], face detection in low-light conditions [14], image denoising in low-light and noisy scenes [15] etc. In the context of IPS, the RSSIs values are collected from a set of pre-deployed beacon devices to form the training dataset at each of the known locations.…”

Automatic and Efficient Metallic Surface Defect Detection Based on Key Pixel Point Locations

“…The main reason is that most defect detection problems need to be resolved by the custom-made solutions. For extracting the discriminative features, in recent years, hand-crafted features methods and deep learning-based methods have emerged and achieved state-of-the-art results [14] [15]. For hand-crafted features methods, Li et al proposed an X-ray-based detection system that extended 2D wavelet transform methods [16].…”

“…Similarity measure functionAt the edge points of the object, the similarity measure function is shown as in(15). Where ' edge gradient vector in the template, (x,y) is the edge point and…”