Human Activity Recognition on Custom Indoor Testbed (test)
69Position AccuracyCNN-ABLSTM with attention
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CNN-ABLSTM with attentionWi-Fi=CSI, 5G=–, Claimed results=97% accuracy, Methodology=CNN-ABLSTM with attention., Limitations of the framework=Person must block the line of sight of the stations.2025.09 | 69 | 86.7 | |
| Modified LeNet-5Wi-Fi=–, 5G=CSI, Claimed results=98.5% accuracy, Methodology=Modified LeNet-5 [14] network., Limitations of the framework=Person must block the line of sight of the stations.2025.09 | 38 | 87.9 | |
| Multi-head attention mechanisms and positional encodingWi-Fi=CSI, 5G=–, Claimed results=98% accuracy, Methodology=Multi-head attention mechanisms and positional encoding., Limitations of the framework=Person must block the line of sight of the stations.2025.09 | 4 | 43.5 |