SlotFM: A Motion Foundation Model with Slot Attention for Diverse Downstream Tasks
This addresses the problem of limited applicability in wearable accelerometer models for researchers and practitioners in fields like gesture recognition and sports monitoring, representing an incremental improvement with a novel method for a known bottleneck.
The paper tackles the limitation of existing accelerometer foundation models that focus on common daily activities by introducing SlotFM, a model that generalizes across diverse downstream tasks using Time-Frequency Slot Attention and achieves a 4.5% average performance gain on 16 tasks.
Wearable accelerometers are used for a wide range of applications, such as gesture recognition, gait analysis, and sports monitoring. Yet most existing foundation models focus primarily on classifying common daily activities such as locomotion and exercise, limiting their applicability to the broader range of tasks that rely on other signal characteristics. We present SlotFM, an accelerometer foundation model that generalizes across diverse downstream tasks. SlotFM uses Time-Frequency Slot Attention, an extension of Slot Attention that processes both time and frequency representations of the raw signals. It generates multiple small embeddings (slots), each capturing different signal components, enabling task-specific heads to focus on the most relevant parts of the data. We also introduce two loss regularizers that capture local structure and frequency patterns, which improve reconstruction of fine-grained details and helps the embeddings preserve task-relevant information. We evaluate SlotFM on 16 classification and regression downstream tasks that extend beyond standard human activity recognition. It outperforms existing self-supervised approaches on 13 of these tasks and achieves comparable results to the best performing approaches on the remaining tasks. On average, our method yields a 4.5% performance gain, demonstrating strong generalization for sensing foundation models.