A Real-Time Deployable Attention-Driven CNN–LSTM Framework for Human Activity Recognition using Wearable Sensor

Human Activity Recognition (HAR) using wearable inertial sensors is increasingly vital for healthcare, rehabilitation, and pervasive computing. Yet, many state-of-the-art models rely on handcrafted feature engineering, heavy time–frequency transformations, or large neural networks, limiting real-time deployment on resource-constrained smartphones and wearables. We propose an end-to-end, lightweight attention-guided CNN–CBAM–LSTM architecture that directly consumes raw IMU signals standardized by StandardScaler, eliminating costly preprocessing. Our design uniquely combines: (i) multi-resolution convolutional kernels to capture diverse motion dynamics; (ii) Convolutional Block Attention Modules (CBAM) to adaptively highlight salient spatial features and suppress sensor noise; and (iii) temporal self-attention atop stacked LSTM layers to model long-range temporal dependencies. Extensive experiments on three public HAR benchmarks—mHealth (controlled), MotionSense (uncontrolled), and UCI HAR (semi-controlled)—show that our model achieves accuracies of 99.7%, 99.5%, and 99.4% respectively, consistently surpassing recent baselines. Moreover, structured pruning and quantization compress the model to just 130.8 KB, while real-world tests on a Snapdragon 845-class Android device confirm ultra-low inference latency of 13.5 ms per segment under live streaming. This fusion of high accuracy, low latency, and minimal memory footprint demonstrates the potential of attention-driven, multi-resolution deep learning for practical, real-time edge AI in HAR applications.