0 views collected around this technical thread.
The paper proposes BCNet, a lightweight bilayer instance segmentation network that explicitly models occluder and occludee relationships by treating each region of interest as two overlapping layers, achieving significant performance gains on COCO, COCOA and KINS datasets under heavy occlusion.
The article introduces BCNet, a lightweight dual‑layer instance segmentation network that models images as overlapping occluder and occludee layers, enabling effective handling of heavy object occlusion and achieving significant performance gains on COCO, COCOA and KINS datasets compared to existing methods.
At the 2020 Yunqi Conference, Amap’s senior algorithm expert presented the Hybrid Spatio‑Temporal Graph Convolutional Network, which leverages massive real‑time navigation data to estimate future traffic flow, transform it into travel‑time features, and outperform prior models, enabling proactive congestion avoidance and dynamic traffic‑scheduling for millions of users.
The Hybrid Spatio‑Temporal Graph Convolutional Network (H‑STGCN) integrates planned traffic flow from navigation data via a domain transformer and a compound adjacency matrix, enabling graph‑based spatio‑temporal modeling that consistently outperforms baselines in real‑world traffic forecasting and reduces severe ETA errors.
JD AI Research Institute presented a GCN‑LSTM encoder‑decoder system that integrates object semantic and spatial relationships via graph convolutional networks to significantly improve image captioning performance on the COCO benchmark, achieving state‑of‑the‑art results.