Human action recognition based on HOIRM feature fusion and AP clustering BOW

Abstract

In this paper, we propose a human action recognition method using HOIRM (histogram of oriented interest region motion) feature fusion and a BOW (bag of words) model based on AP (affinity propagation) clustering. First, a HOIRM feature extraction method based on spatiotemporal interest points ROI is proposed. HOIRM can be regarded as a middle-level feature between local and global features. Then, HOIRM is fused with 3D HOG and 3D HOF local features using a cumulative histogram. The method further improves the robustness of local features to camera view angle and distance variations in complex scenes, which in turn improves the correct rate of action recognition. Finally, a BOW model based on AP clustering is proposed and applied to action classification. It obtains the appropriate visual dictionary capacity and achieves better clustering effect for the joint description of a variety of features. The experimental results demonstrate that by using the fused features with the proposed BOW model, the average recognition rate is 95.75% in the KTH database, and 88.25% in the UCF database, which are both higher than those by using only 3D HOG+3D HOF or HOIRM features. Moreover, the average recognition rate achieved by the proposed method in the two databases is higher than that obtained by other methods.

Complex Human Action Recognition Using a Hierarchical Feature Reduction and Deep Learning-Based Method

“…A confusion matrix contains visualised and quantised information about multiple classifiers using a reference classification system [ 13 , 50 ]. Figure 8 shows the details of the results on the UCF Sports dataset for the proposed HFR-DL method.…”

“…For example, in one of the examined videos “walk-front/006RF1-13902-70016.avi”, there is a person who walks on a golf field with a golf pole. The environment is related to golf field and the motion of the golf pole in the background looks like a person is swinging the pole in front of him [ 13 ]. This was an examples of mis-classifications by the proposed HFR-DL method.…”

“…The feature selection operation is performed in parallel in order to speed up the processing, namely parallel duplex LSTMs. A similar approach is considered in [ 13 , 19 , 50 , 64 ], and [ 29 ]. In other words, we use LSTM for two main reasons: As each frame plays an important role in a video, maintaining the important information of successive frames for a long time will make the system more efficient.…”

Accurate Physical Activity Recognition using Multidimensional Features and Markov Model for Smart Health Fitness

“…UCF Sports Actions dataset (%) Physical Sports Movements [27] 86.67 HOIRM feature fusion [28] 88.25 Hybrid deep learning model [29] 89.01 Proposed method 90.91 A comparison of overall results shows that the proposed method achieved a significant improvement with recognition results as high as 89.09% and 88.26% over other methods as shown in Table 5. Table 5.…”

“…Physical Sports Movements [27] 86.67 HOIRM feature fusion [28] 88.25 Hybrid deep learning model [29] 89.01…”