TY - GEN
T1 - Joint Forecasting of Panoptic Segmentations with Difference Attention
AU - Graber, Colin
AU - Jazra, Cyril
AU - Luo, Wenjie
AU - Gui, Liangyan
AU - Schwing, Alexander
N1 - Weintroduceanewmodelforpanopticsegmentation forecasting. Itusesdifference attentionwhichwefindto bemore suitableto forecastingthan standardattention as itcanreasonaboutvelocitiesandacceleration. Anewre-finementheadalsomergespredictionsbasedondepth. This improves prior work on all panoptic forecasting metrics. Acknowledgements: This work is supported in part byNSF#1718221, 2008387, 2045586, 2106825, MRI #1725729, NIFA2020-67021-32799andCiscoSystems Inc. (CG 1377144 - thanks for access to Arcetri).
This work is supported in part by NSF #1718221, 2008387, 2045586, 2106825, MRI #1725729, NIFA 2020-67021-32799 and Cisco Systems Inc. (CG 1377144 - thanks for access to Arcetri).
PY - 2022
Y1 - 2022
N2 - Forecasting of a representation is important for safe and effective autonomy. For this, panoptic segmentations have been studied as a compelling representation in recent work. However, recent state-of-the-art on panoptic segmentation forecasting suffers from two issues: first, individual object instances are treated independently of each other; second, individual object instance forecasts are merged in a heuristic manner. To address both issues, we study a new panoptic segmentation forecasting model that jointly forecasts all object instances in a scene using a transformer model based on 'difference attention.' It further refines the predictions by taking depth estimates into account. We evaluate the proposed model on the Cityscapes and AIODrive datasets. We find difference attention to be particularly suitable for forecasting because the difference of quantities like locations enables a model to explicitly reason about velocities and acceleration. Because of this, we attain state-of-the-art on panoptic segmentation forecasting metrics.
AB - Forecasting of a representation is important for safe and effective autonomy. For this, panoptic segmentations have been studied as a compelling representation in recent work. However, recent state-of-the-art on panoptic segmentation forecasting suffers from two issues: first, individual object instances are treated independently of each other; second, individual object instance forecasts are merged in a heuristic manner. To address both issues, we study a new panoptic segmentation forecasting model that jointly forecasts all object instances in a scene using a transformer model based on 'difference attention.' It further refines the predictions by taking depth estimates into account. We evaluate the proposed model on the Cityscapes and AIODrive datasets. We find difference attention to be particularly suitable for forecasting because the difference of quantities like locations enables a model to explicitly reason about velocities and acceleration. Because of this, we attain state-of-the-art on panoptic segmentation forecasting metrics.
UR - http://www.scopus.com/inward/record.url?scp=85137843808&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85137843808&partnerID=8YFLogxK
U2 - 10.1109/CVPRW56347.2022.00288
DO - 10.1109/CVPRW56347.2022.00288
M3 - Conference contribution
AN - SCOPUS:85137843808
T3 - IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
SP - 2558
EP - 2567
BT - Proceedings - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2022
PB - IEEE Computer Society
T2 - 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2022
Y2 - 19 June 2022 through 20 June 2022
ER -