Theodore Pissas

Post-doc at the AI in Medical Imaging Lab in ARTORG at @unibern.
PhD from @UCL and research assistant at the @rvimlab at @KingsImaging.

[Theodore Pissas @pis_tho]

@ICCVConference Does removing paragraph indentation count as a significant alteration to the formatting style of the rebuttal? (as described https://t.co/NICEyehJEt)

[Theodore Pissas @pis_tho]

This is a joint work with Claudio Ravasio, Lyndon Da Cruz and @bergeles from UCL and KCL @rvimlab.

[Theodore Pissas @pis_tho]

Happy to share our paper on multi-scale and cross-scale contrastive learning for semantic segmentation was accepted at #ECCV2022. Our paper is on arxiv https://t.co/T2ufwaxLCO. Short thread on the main ideas in the paper:

[Theodore Pissas @pis_tho]

Using both those elements our approach improves performance of various strong baselines comprising both CNNs (DeepLabv3, HRNet, UPerNet, OCRNet) and Transformer-based models (Swin) on both natural (ADE20K, Cityscapes, Pascal-C) and surgical (CaDIS) datasets.