ChatPPG.
Paper

DeepConvContext: A Multi-Scale Approach to Timeseries Classification in Human Activity Recognition

Marius Bock, Juergen Gall, Michael Moeller, Kristof Van Laerhoven
arXiv preprint2026IMUOpen code
Read paper arXiv Code Cite
Published articleFirst posted: May 27, 2025Publication date: Oct 10, 2026Added to atlas: Oct 12, 2026Source
Citations0Crossref referencesChecked Oct 16, 2026GitHub stars4Repository-wideChecked Oct 16, 2026
HF downloads—Not available

Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.

DeepConvContext is a multi-scale activity recognition model that learns across neighbouring windows instead of classifying each window alone. On six common benchmarks it improved F1 by 5% on average and mAP by up to 18 points, with latency and throughput similar to earlier methods. Learning across windows also gave more coherent activity segments in online prediction.

  • Result: Across six widely-used HAR benchmarks, DeepConvContext achieves an average 5% improvement in F1-score and up to 18-point improvement in mAP over related approaches, while achieving latency and throughput comparable to prior methods that extend temporal…
  • Venue: arXiv preprint
  • Signals: IMU
At a glance
Venue
arXiv preprint
Published
2026
DOI
10.1145/3830727.3834810 (opens in a new tab)
arXiv
2505.20894 (opens in a new tab)
PMID
Not reported
Licence
Not reported
Signals
IMU
Form factor
Not reported
Subjects
Not reported
Code
Open (opens in a new tab)
Added
Oct 12, 2026

Abstract

Despite recognized limitations in modeling long-range temporal dependencies, Human Activity Recognition (HAR) has traditionally relied on a sliding window approach to segment labeled datasets. Deep learning models like the DeepConvLSTM typically classify each window independently, restricting learnable temporal context to within-window information and producing fragmented, temporally incoherent activity timelines. To address this constraint, we propose DeepConvContext, a multi-scale time series classification framework for HAR. Drawing inspiration from the vision-based Temporal Action Localization community, DeepConvContext models both intra- and inter-window temporal patterns separately by processing sequences of time-ordered windows. Across six widely-used HAR benchmarks, DeepConvContext achieves an average 5% improvement in F1-score and up to 18-point improvement in mAP over related approaches, while achieving latency and throughput comparable to prior methods that extend temporal context through hidden state propagation across batches. Our quantitative and qualitative analysis underline the importance of inter-window learning and show how it produces more coherent activity segments even in online prediction scenarios. Code to reproduce our experiments is publicly available via www.github.com/mariusbock/deepconvcontext.

Verbatim fromarXiv· arXiv metadata (abstracts are released under CC0)

DeepConvContext: A Multi-Scale Approach to Timeseries Classification in Human Activity Recognition: ChatPPG research summary card

Key results

5%
F1
On six common benchmarks it improved F1 by 5% on average and mAP by up to 18 points, with latency and throughput similar to earlier methods

How it works

How Method works: Setup → Method → ResultSetup: Signals (IMU). Method: Method. Result: Result (Across six widely-used HAR benchmarks, DeepConvContext achieves an average 5% improvement in F1-score and up to 18-point improvement in mAP over related approaches, while achieving latency and throughput comparable to prior methods that extend temporal…).SetupSignals: IMUxyzSignalsIMUMethodMethodResultResult: Across six widely-used HAR benchmarks, DeepConvContext achieves an average 5% improvement in F1-score and up to 18-point improvement in mAP over related approaches, while achieving latency and throughput comparable to prior methods that extend temporal…ResultAcross six widely-used HAR benchmarks…

Cite

@article{deepconvcontextmultiscal2026,
  title = {{DeepConvContext: A Multi-Scale Approach to Timeseries Classification in Human Activity Recognition}},
  author = {Marius Bock and Juergen Gall and Michael Moeller and Kristof Van Laerhoven},
  year = {2026},
  journal = {arXiv preprint},
  doi = {10.1145/3830727.3834810},
  eprint = {2505.20894}
}

History

  1. Added to the ChatPPG research hub

More on ChatPPG