 This paper proposes an end-to-end uncertainty aware finger movement classifier called Evidential Convolutional Neural Network, ECNN. It is designed to improve the reliability of SEMG-based hand gesture recognition. The proposed methodology is evaluated on the NENAPRO Database 5, Exercise A, and compared against other state-of-the-art methods. The results demonstrate that ECNN outperforms existing methods in terms of both accuracy and reliability. Additionally, the proposed reliability analysis provides a new perspective on evaluating the quality of SEMG-based hand gesture recognition models. This article was authored by Yu-Joo Lin, Rameswamy Polanyapan, Philippe De Wilde, and others.