sharp

Description

SHARP is the WiFi CSI sensing system / dataset by Meneghello et al. for environment-robust human activity recognition. It targets the hardware-domain and environment-domain shift problem in CSI sensing and exposes a public dataset of CSI traces collected across multiple devices and rooms so methods can be benchmarked under realistic domain change.

Modality / size

• Modality: WiFi CSI on commodity 802.11ac hardware.
• Subjects / scenarios: multiple environments and devices to evaluate cross-domain robustness.
• Labels: activity class plus environment / device identifier for domain-shift splits.

Used by (papers)

• Cited as a reference for CSI hardware-domain change in the vault.
• Compared against by methods that claim cross-environment / cross-device generalisation.

Notes

• Public release accompanies the SHARP paper.
• This note refers to the WiFi CSI sensing dataset, not the LLM retrieval benchmark also abbreviated SHARP in unrelated literature.