Fall Detection and Activity Recognition Using Human Skeleton Features

Ramirez, Heilym; Velastín, Sergio A.; Meza, Ignacio; Fábregas, Ernesto; Makris, Dimitrios; Farías, Gonzalo

doi:10.1109/access.2021.3061626

Cited by 98 publications

(83 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The present work focuses on improving the performance of fall detection and activity recognition using video data. The main hypothesis is that the results obtained in [ 12 ] can be improved by carrying out a sequential analysis of frames, instead of analyzing and classifying each frame independently, even using the same machine learning models and the same datasets.…”

Section: Methodology Of the Proposed Approachmentioning

confidence: 99%

“…Each window spans an

number of frames, and each frame contains the activity of one person. Following the methodology proposed in [ 12 ], each frame will have the 51 skeleton characteristics associated with the person’s pose and the label associated with the activity performed in that frame. This means that each sliding window will contain

skeletons with their respective associated tag.…”

Section: Methodology Of the Proposed Approachmentioning

confidence: 99%

“…The characteristics to be taken into account are: (a) fall detection, (b) recognition of different activities, and (c) the use of skeletons and a sequential frame analysis methodology, resulting in good performance (accuracy over

). It is important to note that the present work cannot be properly compared with the three mentioned works due to the fact that the training and validation datasets are different (The present work uses the UP-Fall and UR-Fall datasets to compare directly with [ 12 ]). On the other hand, the datasets used in [ 44 , 45 ] (FFD and TST-Fall, respectively) have images with depth information from the use of kinect cameras, which makes the method much easier to apply.…”

Section: Fall Detection Datasets and Related Workmentioning

confidence: 99%

“…In this context, we previously presented [ 12 ] a vision-based approach for fall detection and activity recognition using human skeleton estimation for feature extraction. It can detect several subjects in the same scene, so it uses a filtering method to detect the activity of a single person of interest, discriminating the skeleton of the other subjects.…”

Section: Introductionmentioning

confidence: 99%

“…Regardless of the classification method, in all cases the results are improved. In this approach, the feature extraction method seeks to obtain the pose of the subject to which the activity is identified and for which a skeleton detection algorithm that has delivered excellent results in previous work is used [ 12 ]. This method has been proven with different datasets and machine learning algorithms showing better results than previous works [ 18 , 19 , 20 ].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Human Activity Recognition by Sequences of Skeleton Features

Ramirez

Velastín

Aguayo

et al. 2022

Sensors

Self Cite

View full text Add to dashboard Cite

In recent years, much effort has been devoted to the development of applications capable of detecting different types of human activity. In this field, fall detection is particularly relevant, especially for the elderly. On the one hand, some applications use wearable sensors that are integrated into cell phones, necklaces or smart bracelets to detect sudden movements of the person wearing the device. The main drawback of these types of systems is that these devices must be placed on a person’s body. This is a major drawback because they can be uncomfortable, in addition to the fact that these systems cannot be implemented in open spaces and with unfamiliar people. In contrast, other approaches perform activity recognition from video camera images, which have many advantages over the previous ones since the user is not required to wear the sensors. As a result, these applications can be implemented in open spaces and with unknown people. This paper presents a vision-based algorithm for activity recognition. The main contribution of this work is to use human skeleton pose estimation as a feature extraction method for activity detection in video camera images. The use of this method allows the detection of multiple people’s activities in the same scene. The algorithm is also capable of classifying multi-frame activities, precisely for those that need more than one frame to be detected. The method is evaluated with the public UP-FALL dataset and compared to similar algorithms using the same dataset.

show abstract

Section: Methodology Of the Proposed Approachmentioning

confidence: 99%

“…Each window spans an

skeletons with their respective associated tag.…”

Section: Methodology Of the Proposed Approachmentioning

confidence: 99%