Human Activity Recognition on USC-HAD

70.21Macro F1

TCNet

Updated 1mo ago

Evaluation Results

Method
TCNet 2026.04	70.21	76.21
rTsfNet 2026.04	65.76	64.85
Triplet LSTM 2025.12	62.8	-
SensorLLM 2026.04	61.2	62.6
LightTS 2026.04	59.01	67.78
RAG-HAR 2025.12	58.63	57.2
RandomForest 2026.04	57.61	67.39
Crossformer 2026.04	57.23	67.21
FiLM 2026.04	55.66	64.23
Transformer-like 2025.12	55	-
mobileHART 2026.04	54.54	59.43
TimesNet 2026.04	54.19	60.79
Triplet LSTM 2025.12	53.5	-
iTransformer 2026.04	52.63	58.9
FEDformer 2026.04	51.37	62.91
Transformer 2026.04	49.21	54.42
Softmax LSTM 2025.12	49	-
DeepConvLSTM 2026.04	48.8	50.6
ICGNet 2026.04	46.93	52
Informer 2026.04	46.29	53.67
Pyraformer 2026.04	46.26	51.2
DeepConvLSTM 2025.12	46	-
MchCnnGRU 2026.04	43.07	52.6
Reformer 2026.04	40.33	45.69
CNN+HC 2026.04	-	90.67
CNN-LSTM [12] 2026.04	-	90.88
CNN-LSTM [11] 2026.04	-	90.91
two-level network architecture with dual-stage feature fusion 2026.04	-	94.4