The 2nd International StepUP Competition for Biometric Footstep Recognition: From Steps to Strides

2026-07-15

AI Summary

This work addresses key challenges in plantar pressure–based gait biometrics, including poor generalization, limited robustness to domain shifts induced by varying footwear and walking speeds, and difficulties in fusing left- and right-foot signatures. Leveraging the large-scale StepUP-P150 dataset and a newly curated test set, the study introduces, for the first time, extreme cross-domain conditions and a stride-level verification task. The proposed approach integrates spatiotemporal convolutional networks, multi-stride temporal modeling, ensemble scoring, and inference-time normalization and calibration strategies to significantly enhance model performance. The best-performing system developed by the ArogyaPandit team achieves an equal error rate of 8.00%, demonstrating the efficacy of the proposed methodology while highlighting personalized footwear scenarios as a persistent challenge.

Abstract

The International StepUP Competition Series was launched to advance research in pressure-based footstep biometrics through a standardized and challenging evaluation framework. Using the large-scale StepUP-P150 dataset (with more than 200,000 high-resolution dynamic footsteps from 150 individuals) and a previously unreleased test set, the 2nd edition of the competition addressed three key challenges: (1) generalization to unseen users with limited enrollment data, (2) robustness to domain shift caused by variations in footwear and walking speed and (3) effective fusion of paired left-right footsteps. While the first two challenges built on the inaugural competition, this edition introduced more extreme cross-domain conditions and moved beyond isolated footsteps to stride-level verification, enabling new opportunities for representation learning and inter-step information fusion. The competition attracted 26 registrants from academia and industry, with a best equal error rate of 8.00% achieved by the ArogyaPandit Research Team using a spatiotemporal CNN combined with an ensemble-based scoring strategy. The top solutions showcase the value of harnessing temporal patterns and of incorporating inference-time normalization and calibration strategies to improve scoring. However, the results also reveal that recognizing users in unseen personal footwear remains a challenge, especially in the presence of distractors with similar characteristics.

Problem

• footstep biometrics
• domain shift
• unseen users
• stride-level verification
• left-right foot fusion

Innovation

• stride-level verification
• spatiotemporal CNN
• domain shift robustness
• left-right footstep fusion
• inference-time calibration