International Symposium on Telecommunication (IST)

Home / 11th International Symposium on Telecommunication (IST'2024)

SegLoc: Visual Self-supervised Learning Scheme for Dense Prediction Tasks on X-ray Images

Authors :

Shervin Halat¹ Mohammad Rahmati² Ehsan Nazerfard³

1- Department of Computer Engineering, Amirkabir University of Technology (Polytechnique) 2- Department of Computer Engineering, Amirkabir University of Technology (Polytechnique) 3- Department of Computer Engineering, Amirkabir University of Technology (Polytechnique)

Keywords :

self-supervised learning،contrastive learning،dense prediction task،segmentation،x-ray images

Abstract :

Lately, remarkable advancements of artificial intelligence have been attributed to the integration of self-supervised learning (SSL) scheme. Despite impressive achievements within natural language processing (NLP), SSL in computer vision has not been able to stay on track comparatively. Recently, integration of contrastive learning on top of existing visual SSL models has established considerable progress, thereby being able to outperform supervised counterparts. Nevertheless, the improvements were mostly limited to classification tasks; moreover, few studies have evaluated visual SSL models in real-world scenarios, while the majority considered datasets containing class-wise portrait images, notably ImageNet. Thus, here, we have considered dense prediction tasks on security inspection x-ray images to evaluate our proposed model Segment Localization (SegLoc). Based upon the model Instance Localization (InsLoc), our model has managed to address one of the most challenging downsides of contrastive learning, i.e., false negative pairs of query embeddings. To do so, our pre-training dataset is synthesized by cutting, transforming, then pasting labeled segments, as foregrounds, from an already existing labeled dataset (PIDray) onto instances, as backgrounds, of an unlabeled dataset (SIXray;) further, we fully harness the labels through integration of the notion – one queue per class – into MoCo-v2 memory bank, avoiding false negative pairs. Regarding the task in question, our approach has outperformed random initialization method by 3% to 6%, while having underperformed supervised initialization, in AR and AP metrics at different IoU values for 20 to 30 pre-training epochs.