Search papers, labs, and topics across Lattice.
TDFNet introduces a novel Tri-projection Deformable Fusion Network specifically designed for panoramic salient object detection, addressing the geometric distortions inherent in existing projection methods like Equirectangular and cube map projections. By employing a cross-projection deformable attention module, TDFNet enhances feature discriminability and geometric consistency through geometry-aware sampling, while a latitude-guided fusion module balances features from different projections. The architecture effectively integrates three projection representations, resulting in significant improvements in detection performance and robustness against distortions.
TDFNet achieves unprecedented robustness in panoramic object detection by integrating three distinct projection methods to counteract geometric distortions.
Recent years have witnessed the growing potential of panoramic salient object detection in robotic vision, virtual reality, and related applications. However, projecting spherical scenes onto 2D planes inevitably introduces geometric distortions, which fundamentally limit the effectiveness of existing projection-based methods. Specifically, Equirectangular Projection (ERP) suffers from severe polar stretching distortions, while cube map projection introduces discontinuities across cube-face boundaries, resulting in degraded feature discriminability and compromised geometric consistency. To address these limitations, we propose TDFNet, the first Tri-projection Deformable Fusion Network for panoramic salient object detection, exploiting complementary projection representations to alleviate geometric distortions and improve detection performance.Specifically, we design a cross-projection deformable attention (CDA) module that leverages spatial correspondences between different projections to construct geometry-aware sampling locations, guiding deformable attention for cross-projection contextual aggregation and enhancing robustness against projection-induced deformations. Furthermore, we introduce a latitude-guided fusion module, which utilizes spherical latitude priors to construct geometric confidence weights for adaptively balancing ERP and CMP features. Meanwhile, LGF incorporates distortion-reduced semantic references from Tangent Projection to achieve cross-projection feature refinement and spatial alignment.By constructing a three-branch encoding architecture based on ERP, CMP, and Tangent Projection, TDFNet simultaneously preserves global spatial continuity, local geometric details, and fine-grained boundary information.
