Porovnať metódy

	Multimodal Instance Segmentation	Inštancová segmentácia
Odbor	Hlboké učenie	Hlboké učenie
Rodina	Machine learning	Machine learning
Rok vzniku	2017–present	2017
Tvorca	He, K., Gkioxari, G., Dollar, P., Girshick, R. (Mask R-CNN foundation); extended by community to multimodal settings	He, K., Gkioxari, G., Dollar, P., Girshick, R.
Typ	Supervised deep learning — instance segmentation	Pixel-level detection and mask prediction
Pôvodný zdroj	He, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017). Mask R-CNN. Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2961–2969.	He, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017). Mask R-CNN. Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2961–2969.
Ďalšie názvy	multimodal Mask R-CNN, RGB-D instance segmentation, multi-sensor instance segmentation, cross-modal instance segmentation	instance-level segmentation, object instance segmentation, mask prediction, panoptic instance segmentation
Príbuzné	5	4
Zhrnutie	Multimodal instance segmentation extends classical instance segmentation — which assigns a per-pixel mask and a class label to every individual object in an image — by incorporating complementary sensor streams such as depth maps, LiDAR point clouds, or infrared frames. Fusing these modalities helps the model handle ambiguous appearances, low light, and occlusion that trip up RGB-only systems.	Instance segmentation is a computer vision task that simultaneously detects every distinct object in an image and produces a precise pixel-level mask for each individual object instance. Unlike semantic segmentation, which labels every pixel with a class, instance segmentation distinguishes between separate objects of the same class, enabling fine-grained spatial understanding.