Attend and Discriminate

Abedin, Alireza; Ehsanpour, Mahsa; Shi, Qinfeng; Rezatofighi, Hamid; Ranasinghe, Damith C.

doi:10.1145/3448083

Cited by 47 publications

(13 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our experiment, we quantitatively evaluate the model from two aspects of segmentation and recognition. We employ the accuracy (Acc), class-average F-score (F m ) as the recognition metrics and the Jaccard Index (J I) [55], Intersection over Union (IoU ) [18], and Overfill/Underfill (O/U ) [56] as the segmentation metrics. Calculate L cls , L seg , L con , L s and L conf with Eq.…”

Section: Evaluation Protocols and Datasetsmentioning

confidence: 99%

“…• A&D [56]: A novel deep learning method for HAR, which contains the CIE, AGE, and center-loss module. And it has obtained the SOTA recognition performance through a data enhancement approach.…”

Section: Comparison With the State-of-the-artmentioning

confidence: 99%

“…• Skoda Dataset [60] targets a manufacturing scenario and includes 10 manipulative gestures. For our hold-out evaluation, we split the dataset into training, validation, and testing sets at a ratio of 8:1:1, following [56].…”

mentioning

confidence: 99%

“…Table 4 provides a comprehensive comparison of our proposed method with other techniques under both fully and weakly supervised settings. For a fair comparison, we use the samplewise evaluation protocol [9], [56] for all comparing methods. Despite the fact that our method is primarily designed for a weakly supervised setting, it remains competitive under a fully supervised setting.…”

mentioning

confidence: 99%

See 3 more Smart Citations

A Boundary Consistency-Aware Multitask Learning Framework for Joint Activity Segmentation and Recognition With Wearable Sensors

Xia

Chu

Pei

et al. 2023

IEEE Trans. Ind. Inf.

View full text Add to dashboard Cite

Human activity recognition (HAR) with wearables is one of the serviceable technologies in ubiquitous and mobile computing applications. The sliding-window scheme is widely adopted while suffering from the multi-class windows problem. As a result, there is a growing focus on joint segmentation and recognition with deep-learning methods, aiming at simultaneously dealing with HAR and time-series segmentation issues. However, obtaining the full activity annotations of wearable data sequences is resource-intensive or time-consuming, while unsupervised methods yield poor performance. To address these challenges, we propose a novel method for joint activity segmentation and recognition with timestamp supervision, in which only a single annotated sample is needed in each activity segment. However, the limited information of sparse annotations exacerbates the gap between recognition and segmentation tasks, leading to sub-optimal model performance. Therefore, the prototypes are estimated by class-activation maps to form a sample-to-prototype contrast module for well-structured embeddings. Moreover, with the optimal transport theory, our approach generates the sample-level pseudo-labels that take advantage of unlabeled data between timestamp annotations for further performance improvement. Comprehensive experiments on four public HAR datasets demonstrate that our model trained with timestamp supervision is superior to the state-of-the-art weakly-supervised methods and achieves comparable performance to the fully-supervised approaches.

show abstract

Section: Evaluation Protocols and Datasetsmentioning

confidence: 99%

Section: Comparison With the State-of-the-artmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

A Boundary Consistency-Aware Multitask Learning Framework for Joint Activity Segmentation and Recognition With Wearable Sensors

Xia

Chu

Pei

et al. 2023

IEEE Trans. Ind. Inf.

View full text Add to dashboard Cite

show abstract

“…We investigated a variety of prediction scenarios to provide a first impression of potential test cases and benchmark scores that can be achieved using the Hang-Time HAR dataset. As architectures for our classifiers, we chose to use both a shallow variant of the DeepConvLSTM network [102] and Attend-and-Discriminate network [103]. Each of the defined training scenarios employs either a Leave-One-Subject-Out (LOSO) cross-validation or a train-test split to evaluate a network's predictive performance.…”

Section: Deep Learning Analysismentioning

confidence: 99%

Hang-Time HAR: A Benchmark Dataset for Basketball Activity Recognition Using Wrist-Worn Inertial Sensors

Hoelzemann

Romero

Bock

et al. 2023

Sensors

View full text Add to dashboard Cite

We present a benchmark dataset for evaluating physical human activity recognition methods from wrist-worn sensors, for the specific setting of basketball training, drills, and games. Basketball activities lend themselves well for measurement by wrist-worn inertial sensors, and systems that are able to detect such sport-relevant activities could be used in applications of game analysis, guided training, and personal physical activity tracking. The dataset was recorded from two teams in separate countries (USA and Germany) with a total of 24 players who wore an inertial sensor on their wrist, during both a repetitive basketball training session and a game. Particular features of this dataset include an inherent variance through cultural differences in game rules and styles as the data was recorded in two countries, as well as different sport skill levels since the participants were heterogeneous in terms of prior basketball experience. We illustrate the dataset’s features in several time-series analyses and report on a baseline classification performance study with two state-of-the-art deep learning architectures.

show abstract