Efficient Adaptation of Deep Neural Networks for Semantic Segmentation in Space Applications
This work addresses the challenge of deploying deep learning models in extraterrestrial environments with limited labeled data and device constraints, representing an incremental advancement in transfer learning methods for space applications.
This paper tackles the problem of efficiently adapting deep neural networks for semantic segmentation in space applications, such as lunar and martian terrains, by using adapters to reduce bandwidth and memory requirements, achieving results that demonstrate trade-offs in task performance, memory, and computation on embedded devices.
In recent years, the application of Deep Learning techniques has shown remarkable success in various computer vision tasks, paving the way for their deployment in extraterrestrial exploration. Transfer learning has emerged as a powerful strategy for addressing the scarcity of labeled data in these novel environments. This paper represents one of the first efforts in evaluating the feasibility of employing adapters toward efficient transfer learning for rock segmentation in extraterrestrial landscapes, mainly focusing on lunar and martian terrains. Our work suggests that the use of adapters, strategically integrated into a pre-trained backbone model, can be successful in reducing both bandwidth and memory requirements for the target extraterrestrial device. In this study, we considered two memory-saving strategies: layer fusion (to reduce to zero the inference overhead) and an ``adapter ranking'' (to also reduce the transmission cost). Finally, we evaluate these results in terms of task performance, memory, and computation on embedded devices, evidencing trade-offs that open the road to more research in the field.