Human Activity Recognition on UCI HAR (Performance and Efficiency)
0.21Inference latency (ms)SPECTRA
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SPECTRAPrecision=FP162026.03 | 0.21 | 1,677.5 | — | — | — | — | 93.3 | 10,812 | 113,112 | 17.6 | 0.955 | 0.0001 | |
| SPECTRAPrecision=INT82026.03 | 0.33 | 1,040.3 | — | — | — | — | 85.5 | 10,812 | 113,112 | 17.6 | 1.238 | 0.0001 | |
| SPECTRAPrecision=FP322026.03 | 0.99 | 353.7 | — | — | — | — | 94.8 | 10,812 | 113,112 | 38.2 | 29.5 | 0.0012 | |
| SPECTRAPrecision=INT82026.03 | 2.1 | 476.2 | — | — | — | — | — | — | — | 30.8 | 1.3741 | 0.0045 | |
| SPECTRAPrecision=INT82026.03 | 2.488 | 401.9 | 8.4 | 3.29 | 99.747 | 0.33 | — | — | — | — | — | — | |
| SPECTRAPrecision=FP162026.03 | 2.5 | 392.2 | — | — | — | — | — | — | — | 34.6 | 1.6685 | 0.0055 | |
| TinierHAR2026.03 | 2.68 | 373.1 | — | — | — | — | 95.1 | 9,589 | 637,414 | 55.4 | 60.3 | 0.0024 | |
| SPECTRAPrecision=FP322026.03 | 3 | 333.3 | — | — | — | — | — | — | — | 38.5 | 1.963 | 0.0065 | |
| TinierHAR2026.03 | 3.1 | 322.6 | — | — | — | — | — | — | — | 77 | 2.2952 | 0.0076 | |
| SPECTRAPrecision=FP322026.03 | 3.317 | 301.4 | 23.99 | 3.29 | 166.245 | 0.55 | — | — | — | — | — | — | |
| TinyHAR2026.03 | 11.171 | 89.5 | 114.78 | 106.36 | 195,170.129 | 646.1 | — | — | — | — | — | — | |
| TinyHAR2026.03 | 11.71 | 85.4 | — | — | — | — | 95.3 | 29,828 | 3,859,938 | 99.6 | 199.9 | 0.0078 | |
| TinyHAR2026.03 | 12.9 | 77.5 | — | — | — | — | — | — | — | 173.3 | 9.3922 | 0.0311 | |
| DeepConvLSTM2026.03 | 35.55 | 28.1 | — | — | — | — | 95.5 | 424,262 | 36,422,534 | 226.9 | 554.3 | 0.0217 | |
| DeepConvLSTM2026.03 | 36.82 | 27.2 | 1,331.05 | 5.04 | 563,623.27 | 1,866.9 | — | — | — | — | — | — | |
| DConvLSTM2026.03 | 37.5 | 26.7 | — | — | — | — | — | — | — | 1,231 | 23.9788 | 0.0794 |