We propose a novel, conceptually simple and general framework for instance segmentation on 3D point clouds. Our method, called 3D-BoNet, follows the simple design philosophy of per-point multilayer perceptrons (MLPs). The framework directly regresses
3D bounding boxes for all instances in a point cloud, while simultaneously predicting a point-level mask for each instance. It consists of a backbone network followed by two parallel network branches for 1) bounding box regression and 2) point mask prediction. 3D-BoNet is single-stage, anchor-free and end-to-end trainable. Moreover, it is remarkably computationally efficient as, unlike existing approaches, it does not require any post-processing steps such as non-maximum suppression, feature sampling, clustering or voting. Extensive experiments show that our approach surpasses existing work on both ScanNet and S3DIS datasets while being approximately 10x more computationally efficient. Comprehensive ablation studies demonstrate the effectiveness of our design.
Pauli blockade occurs when the excited electrons fill up the states near the conduction bands and block subsequent absorption in semiconductors, and has been widely applied in mode-locking for passively-pulsed-laser systems. In this letter, we report
the first direct observation that the Pauli blockade is broken by ultrafast cooling of hot electrons in optically-pumped graphene. With femtosecond spectroscopy, we demonstrate that the time scale to excite an electron (~100 fs) is of the same order as that of the electron decay via electron-electron scattering, which allows the electron excitation interplays strongly with the cooling of hot electrons. Consequently, Pauli blockade is dismissed, leading to an unconventionally enhanced optical absorption. We suggest that this effect is a universal feature of two-dimensional layered materials, which sheds the light of ultrafast carrier dynamics in nonlinear physics and inspires the designing of new-generation of ultrafast optoelectronic devices.
