Weakly- and semi-supervised panoptic segmentation by Li, Q, Arnab, A, Torr, P

“…Moreover, we are able to segment both “thing” and “stuff” classes, and thus explain all the pixels in the image. …”

Unifying training and inference for panoptic segmentation by Li, Q, Qi, X, Torr, PHS

“…This panoptic submodule gives rise to a novel propagation mechanism for panoptic logits and enables the network to output a coherent panoptic segmentation map for both "stuff" and "thing" classes, without any post-processing. …”

‘Whether any material being thinks, or no’: Leibniz’s critique of Locke on superaddition by Lodge, P

“…The mill argument aims to show that matter conceived as passive extended stuff cannot think through its natural powers. This is something Locke is happy to accept, noting explicitly at ECHU 4.3.6: “matter …is evidently in its own nature void of sense and thought”. …”

Open vocabulary semantic segmentation with Patch Aligned Contrastive Learning by Mukhoti, J, Lin, T-Y, Poursaeed, O, Wang, R, Shah, A, Torr, PHS, Lim, S-N

“…Using pre-trained CLIP encoders with PACL, we are able to set the state-of-the-art on the task of open vocabulary zero-shot segmentation on 4 different segmentation benchmarks: Pascal VOC, Pascal Context, COCO Stuff and ADE20K. Furthermore, we show that PACL is also applicable to image-level predictions and when used with a CLIP backbone, provides a general improvement in zero-shot classification accuracy compared to CLIP, across a suite of 12 image classification datasets.…”