Online Detection of Long-Term Daily Living Activities by Weakly Supervised Recognition of Sub-Activities

Negin, Farhood; Goel, Abhishek; Abubakr, Abdelrahman G.; Brémond, François; Francesca, Gianpiero

doi:10.1109/avss.2018.8639471

Cited by 3 publications

(6 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Inspired by the Hough transformation that is successfully applied in object detection, Chan-Hon-Tong et al [91] proposes a method with discriminative features to globally optimize the parameters of Hough transform and utilize it for activity segmentation in videos. Finally, our results are compared with [92] that is a supervised method with a semi-supervised component to discover sub-activities. Table 5 demonstrates our results on the DAHLIA dataset.…”

Section: Experiments and Discussionmentioning

confidence: 99%

An Unsupervised Framework for Online Spatiotemporal Detection of Activities of Daily Living by Hierarchical Activity Models

Negin

Brémond

2019

Sensors

Self Cite

View full text Add to dashboard Cite

Automatic detection and analysis of human activities captured by various sensors (e.g., sequences of images captured by RGB camera) play an essential role in various research fields in order to understand the semantic content of a captured scene. The main focus of the earlier studies has been widely on supervised classification problem, where a label is assigned to a given short clip. Nevertheless, in real-world scenarios, such as in Activities of Daily Living (ADL), the challenge is to automatically browse long-term (days and weeks) stream of videos to identify segments with semantics corresponding to the model activities and their temporal boundaries. This paper proposes an unsupervised solution to address this problem by generating hierarchical models that combine global trajectory information with local dynamics of the human body. Global information helps in modeling the spatiotemporal evolution of long-term activities, hence, their spatial and temporal localization. Moreover, the local dynamic information incorporates complex local motion patterns of daily activities into the models. Our proposed method is evaluated using realistic datasets captured from observation rooms in hospitals and nursing homes. The experimental data on a variety of monitoring scenarios in hospital settings reveals how this framework can be exploited to provide timely diagnose and medical interventions for cognitive disorders, such as Alzheimer’s disease. The obtained results show that our framework is a promising attempt capable of generating activity models without any supervision.

show abstract

Section: Experiments and Discussionmentioning

confidence: 99%

An Unsupervised Framework for Online Spatiotemporal Detection of Activities of Daily Living by Hierarchical Activity Models

Negin

Brémond

2019

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…Each frame codeword has a certain weight in the vote for assigning the label of neighboring frames, and the weighting function is learned using a novel optimization method. Negin et al [16] utilize a sliding window approach. This method obtains features from Resnet-152 [9] for each frame to form a code-book and then trains a SVM classifier.…”

Section: State-of-the-art Methodsmentioning

confidence: 99%

“…The mean duration of activities is 6 mins. We performed experiments using the cross-subject protocol described in [16]. The final result is obtained as the average of the results on the 3 camera views.…”

Section: Datasetsmentioning

confidence: 99%

“…Some existing state-of-the-art methods address this task by using sliding window approaches [31,16]. Generally, these methods use a classifier trained on trimmed videos.…”

Section: Introductionmentioning

confidence: 99%

“…Detection visualization. The detection visualization of video 'S01A2K1' in DAHLIA: (1) ground truth, (2) GRU[18], (3) ED-TCN[13], (4) TCFPN[16] and (5) SA-TCN.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Self-Attention Temporal Convolutional Network for Long-Term Daily Living Activity Detection

Dai

Minciullo

Garattoni

et al. 2019

2019 16th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)

Self Cite

View full text Add to dashboard Cite

In this paper, we address the detection of daily living activities in long-term untrimmed videos. The detection of daily living activities is challenging due to their long temporal components, low inter-class variation and high intra-class variation. To tackle these challenges, recent approaches based on Temporal Convolutional Networks (TCNs) have been proposed. Such methods can capture long-term temporal patterns using a hierarchy of temporal convolutional filters, pooling and up sampling steps. However, as one of the important features of convolutional networks, TCNs process a local neighborhood across time which leads to inefficiency in modeling the long-range dependencies between these temporal patterns of the video. In this paper, we propose Self-Attention-Temporal Convolutional Network (SA-TCN), which is able to capture both complex activity patterns and their dependencies within long-term untrimmed videos. We evaluate our proposed model on DAily Home LIfe Activity Dataset (DAHLIA) and Breakfast datasets. Our proposed method achieves state-of-the-art performance on both DAHLIA and Breakfast dataset.

show abstract

TN-ZSTAD: Transferable Network for Zero-Shot Temporal Activity Detection

Zhang

Chang

Liu

et al. 2022

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

Online Detection of Long-Term Daily Living Activities by Weakly Supervised Recognition of Sub-Activities

Cited by 3 publications

References 15 publications

An Unsupervised Framework for Online Spatiotemporal Detection of Activities of Daily Living by Hierarchical Activity Models

An Unsupervised Framework for Online Spatiotemporal Detection of Activities of Daily Living by Hierarchical Activity Models

Self-Attention Temporal Convolutional Network for Long-Term Daily Living Activity Detection

TN-ZSTAD: Transferable Network for Zero-Shot Temporal Activity Detection

Contact Info

Product

Resources

About