Face recognition under extreme head poses is a challenging task. Ideally, a face recognition system should perform well across different head poses, which is known as pose-invariant face recognition. To achieve pose invariance, current approaches rely on sophisticated methods, such as face frontalization and various facial feature extraction model architectures. However, these methods are somewhat impractical in real-life settings and are typically evaluated on small scientific datasets, such as Multi-PIE. In this work, we propose the inverse method of face frontalization, called face defrontalization, to augment the training dataset of facial feature extraction model. The method does not introduce any time overhead during the inference step. The method is composed of: 1) training an adapted face defrontalization FFWM model on a frontal-profile pairs dataset, which has been preprocessed using our proposed face alignment method; 2) training a ResNet-50 facial feature extraction model based on ArcFace loss on a raw and randomly defrontalized large-scale dataset, where defrontalization was performed with our previously trained face defrontalization model. Our method was compared with the existing approaches on four open-access datasets: LFW, AgeDB, CFP, and Multi-PIE. Defrontalization shows improved results compared to models without defrontalization, while the proposed adjustments show clear superiority over the state-of-the-art face frontalization FFWM method on three larger open-access datasets, but not on the small Multi-PIE dataset for extreme poses (75 and 90 degrees). The results suggest that at least some of the current methods may be overfitted to small datasets.

Context. The Spectrometer/Telescope for Imaging X-Rays (STIX) onboard Solar Orbiter was designed to observe solar flares in the X-ray range of 4-150 keV, providing spectral, temporal and spatial information. Besides 30 imaging detectors, STIX has two additional detectors, the coarse flare locator (CFL) and the background (BKG) detector. Flares observed from Earth are classified using their peak X-ray flux observed by the GOES satellites. Roughly half of all flares observed by STIX are located on the backside of the Sun. These flares lack a GOES-class classification. Aims. In this paper, we describe the calibration of the BKG detector aperture sizes. Using the calibrated measurements of the BKG detector, we explore the relationship between the peak flux for flares jointly observed by STIX and GOES. This allows us to estimate the GOES flare classes of backside flares using STIX measurements. Methods. We looked at the 500 largest flares observed by both STIX and GOES in the time range Feb. 21 to Apr. 23. Aperture size calibration is done by comparing 4-10 keV counts of the BKG detector with the CFL measurements. In a second step, we correlate the calibrated STIX BKG peak flux with the GOES peak flux for individual flares. Results. We calibrated the BKG detector aperture sizes of STIX. Further, we showed that for the larger flares a close power law fit exists between the STIX BKG and GOES peak flux with a Pearson correlation coefficient of 0.97. This correlation provides a GOES proxy with a one sigma uncertainty of 11%. We were able to show that the BKG detector can reliably measure a broad range of GOES flare classes from roughly B5 up to at least X85 (assuming a radial distance of 1AU), making it an interesting detector-concept for future space weather missions. The largest flare observed by STIX to date is an estimated X16.5 $\pm$ 1.8 backside flare on the 20 Mai 2024.