Qosja, DenisaDenisaQosjaWagner, SimonSimonWagnerBrüggenwirth, StefanStefanBrüggenwirth2023-09-082023-09-082023https://publica.fraunhofer.de/handle/publica/45042810.1109/RadarConf2351548.2023.101498022-s2.0-85163720807With the recent developments in the field of deep learning, various neural networks have been proposed to increase the detection accuracy of targets in radar data and beyond. A prominent network, named ConvNeXt has achieved state-of-the-art results in computer vision. In this paper, its performance on SAR is aimed to be evaluated and compared to its predecessors over three distinct SAR datasets. A thorough comparison shows the superiority of ConvNeXt in the target recognition task. Furthermore, several augmentations are exploited to enhance the size of training set and evaluated to show whether they fit in the radar domain.enATRCNNConvNeXtdata augmentationtarget classificationconference paper