In the past decade, Deep Learning (DL) became the dominant trend in image data analysis, mostly due to the capacity of DL models to learn discriminative features directly from data, when labeled samples are abundant.
At the same time, the availability of Earth observation (EO) data produced by RS systems has increased considerably. However, most of the RS applications still fall short in the demands imposed by DL-based techniques, basically because of the high costs required by field survey and labor-intensive visual interpretation to produce a large enough quantity of labeled data. The development of wide-reaching DL-based solutions for EO problems, therefore, remains a challenging problem.
In this sense, transfer learning is an attractive alternative, allowing the reuse of networks already trained on large data-sets in problems in which a limited quantity of labeled data is available. Such techniques, however, perform poorly when the domain shift phenomenon is significant. Considering EO applications, changes in the environmental conditions during the acquisition of new data, variations of objects’ appearances, geographical variability, and different sensor properties, domain shift makes it impossible to employ even fine-tuned classifiers over new data without a significant decrease in classification accuracy
Domain adaptation techniques can be used to alleviate the domain shift problem. In short, domain adaptation aims at minimizing the discrepancy between distributions of two different domains. One of the distributions characterizes the data used to train a classifier; the other is associated with data that the classifier has never seen, which may present several of the aforementioned variations.
This SBSR Thematic Session aims at describing and discussing some state-of-the-art Domain Adaptation techniques applied to Earth observation data, such as feature adaptation and image translation.
