方法证据记录
Multimodal Instance Segmentation
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.
源记录
引文逐字复制自方法源记录。这些引文不代表任何层级的验证。
Multimodal Instance Segmentation (Multi-sensor Deep Mask Prediction)
分类方法记录 · ml-model / deep-learning
- He, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017). Mask R-CNN. Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2961–2969. · DOI 10.1109/ICCV.2017.322
- Instance segmentation. Wikipedia. · URL
精选声明
声明已持久化到证据分类账中，每个声明都有自己的评估。
尚无精选声明
当分类账中没有声明时，此视图不会自行创建声明评估。
相关方法
从方法图中生成，显示为机器建议的关系 — 不推断任何证据声明。