Effective Video Scene Analysis for a Nanosatellite Based on an Onboard Deep Learning Method

The latest advancements in satellite technology have allowed us to obtain video imagery from satellites. Nanosatellites are becoming widely used for earth-observing missions as they require a low budget and short development time. Thus, there is real interest using nanosatellites with payload camera, especially disaster monitoring fleet tracking. However, data requires much storage high communication costs, it challenging use such missions. This paper proposes an effective onboard deep-learning-based scene analysis method reduce the cost. proposed will train CNN+LSTM-based model identify mission-related sceneries flood-disaster-related scenery videos on ground then load nanosatellite perform before sending ground. We experimented Nvidia Jetson TX2 OBC achieved 89% test accuracy. Additionally, by implementing our approach, we can minimize download cost 30% which allows send important mission S-band communication. Therefore, believe that new approach be effectively applied large nanosatellite.