Application for semantic segmentation models in areas such as autonomous vehicles and human computer interaction require real-time predictive capabilities. The challenges of addressing real-time application is amplified by the need to operate on resource constrained hardware. Whilst development of real-time methods for these platforms has increased, these models are unable to sufficiently reason about uncertainty present. This paper addresses this by combining deep feature extraction from pre-trained models with Bayesian regression and moment propagation for uncertainty aware predictions. We demonstrate how the proposed method can yield meaningful uncertainty on embedded hardware in real-time whilst maintaining predictive performance.

翻译：在诸如自主车辆和人类计算机互动等领域应用语义分解模型需要实时预测能力。处理实时应用的挑战因需要使用资源有限的硬件而更加艰巨。虽然这些平台实时方法的开发有所增加,但这些模型无法充分说明目前存在的不确定性。本文件通过将预先培训的模型的深度特征提取与贝叶斯回归和瞬间传播相结合,以了解不确定性的预测来解决这一问题。我们展示了拟议方法如何能够在实时嵌入的硬件上产生有意义的不确定性,同时保持预测性能。