IBGS: A Wearable Smart System to Assist Visually Challenged

Traditional blind guide devices are large and high cost. In this paper, a kind of intelligent system (IBGS) is introduced. GD32 used as the main control chip, it cooperates with various functional modules to realize traffic light recognition, obstacle avoidance, payment, navigation functions on basis speech recognition. At same time, IBGS uses WIFI instead Bluetooth get rid dependence smart phones. addition, cloud database built, Internet Things technology information interconnection between IBGS, database, guardian’s mobile terminal. order better function paper proposes Conv-Transformer Transducer (ConvT-T) recognition framework based Weak-Attention Suppression (WAS), which improves efficiency multi-head attention through WAS. The proposed method achieves word error rate (WER) 3.2% in test-clean 7.9% test-other LibriSpeech ASR corpus only 73M parameters, reduces WER by 0.3% 0.4% compared ConvT-T respectively, indicating that WAS can effectively play role suppressing non-critical framework. has been tested comprehensive application scenario. outdoor avoidance tests, accuracy rates 92.33%, 90.33%, 96.67% were achieved respectively. results show deal challenges encountered their daily walking.