NA-SODINN: A deep learning algorithm for exoplanet image detection based on residual noise regimes

Context. Supervised deep learning was recently introduced in high-contrast imaging (HCI) through the SODINN algorithm, a convolutional neural network designed for exoplanet detection in angular differential imaging (ADI) datasets. The benchmarking of HCI algorithms within the Exoplanet Imaging Data Challenge (EIDC) showed that (i) SODINN can produce a high number of false positives in the final detection maps, and (ii) algorithms processing images in a more local manner perform better. Aims. This work aims to improve the SODINN detection performance by introducing new local processing approaches and adapting its learning process accordingly. Methods. We propose NA-SODINN, a new deep learning binary classifier based on a convolutional neural network (CNN) that better captures image noise correlations in ADI-processed frames by identifying noise regimes. The identification of these noise regimes is based on a novel technique, named PCA-pmaps, which allowed us to estimate the distance from the star in the image from which background noise started to dominate over residual speckle noise. NA-SODINN was also fed with local discriminators, such as signal-to-noise ratio (S/N) curves, which complement spatio-temporal feature maps during the model's training. Results. Our new approach was tested against its predecessor, as well as two SODINN-based hybrid models and a more standard annular-PCA approach, through local receiving operating characteristics (ROC) analysis of ADI sequences from the VLT/SPHERE and Keck/NIRC-2 instruments. Results show that NA-SODINN enhances SODINN in both sensitivity and specificity, especially in the speckle-dominated noise regime. NA-SODINN is also benchmarked against the complete set of submitted detection algorithms in EIDC, in which we show that its final detection score matches or outperforms the most powerful detection algorithms. Conclusions. Throughout the supervised machine learning case, this study illustrates and reinforces the importance of adapting the task of detection to the local content of processed images.