Exploiting In-Sensor Computing for Energy-Efficient Earth Observation
Researchers have developed a new in-sensor computing framework for energy-efficient Earth observation from satellites. This approach integrates TinyML techniques with the Sony IMX500 Intelligent Vision Sensor to process data directly on the sensor, reducing the need to transmit raw data. The system achieves 96.68% accuracy on the EuroSAT dataset while operating within the sensor's memory constraints and demonstrating high energy efficiency. AI
IMPACT Enables more efficient satellite data processing and transmission by moving computation to the sensor.