The field of 3D object detection and reconstruction is rapidly advancing, with a focus on improving the robustness and accuracy of detection methods in compressed point clouds and scattering media. Researchers are exploring innovative approaches, such as reflectance prediction-based knowledge distillation and physics-guided mixture-of-experts frameworks, to address the challenges posed by limited bandwidth and complex signal propagation. Additionally, there is a growing interest in leveraging neural radiance fields to infer environmental information from multipath signals, enabling applications such as indoor floorplan reconstruction and signal prediction. Noteworthy papers in this area include:

• EvidenceMoE, which presents a physics-guided framework for advancing fluorescence light detection and ranging in scattering media, achieving strong performance in non-invasive cancer cell depth detection.
• SpatialSplat, a feedforward framework that produces redundancy-aware Gaussians and captures fine-grained semantics, making semantic 3D reconstruction more applicable.
• PhysicsNeRF, a physically grounded framework for 3D reconstruction from sparse views, extending neural radiance fields with complementary constraints to achieve physically consistent and generalizable representations.