Subjects: Computer Science >> Other Disciplines of Computer Science submitted time 2023-12-05
Abstract： Human parsing is a fundamental task aimed at segmenting human images into distinct body parts and holds vast potential applications. Nowadays, the advancement of image-capturing devices has led to a growing number of high-resolution human images. Receptive field, details loss and memory usage are a triplet of contradictions in high-resolution scenarios. Existing human parsing methods designed for low-resolution inputs struggle to process high-resolution images efficiently due to their massive demands for computation and memory. Some methods save resources by overwhelmingly downsampling or encoding high-resolution inputs at the cost of poor performance on details. To resolve the issues above, we propose the Bilateral Edge-Perceiving Network (BiEPNet), consisting of a resources-friendly semantic-perceiving branch to acquire sufficient global information and a simple yet effective edge-perceiving branch used to refine details. The attention mechanism is utilized to simultaneously enhance the perception of context and details, leading to better performance on the boundary regions. To verify the effectiveness of BiEPNet, we contribute a high-resolution human parsing dataset, Human4K, containing 4,000 images with more than five million pixels. Extensive experiments on Human4K demonstrate that our method outperforms state-of-the-art methods while maintaining memory efficiency.
Peer Review Status:
Awaiting Review
Subjects: Optics >> Quantum optics submitted time 2023-02-19
Abstract： Although tremendous achievements have been made toward inertial confinement fusion, laser plasma instabilities (LPIs) remain to be an inevitable problem for current drive schemes. To mitigate these instabilities, significant efforts have been paid to produce high-power broadband ultraviolet lasers. However, no practical scheme has been demonstrated up to now for efficient triple-frequency conversion of broadband laser. Here we propose the design of polychromatic drivers for the generation of multicolor beams mainly based upon the optical parametric amplification, which can significantly enhance the third-harmonic conversion efficiency. Each polychromatic light has four colors of monochromatic beamlets with a full spectrum width of 3\%, and the beamlet colors of any two adjacent flanges are different. The suppression effects of such polychromatic lights have been investigated via large scale particle-in-cell simulations, which indicate that more than 35\% of the incident energy can be saved from the LPIs compared with monochromatic lasers for the direct-drive scheme, or high-density filled target for the indirect-drive scheme. The proposed polychromatic drivers are based on the matured technologies, and thus may pave the way towards realization of robust and high-efficiency fusion ignition.
Peer Review Status:Awaiting Review
Subjects: Optics >> Quantum optics submitted time 2023-02-19
Abstract： Stimulated Raman scattering (SRS) in plasma in the degenerate regime is studied theoretically and numerically. Different from normal SRS with the non-degenerate eigen electrostatic mode excited, the degenerate SRS is developed at plasma density $n_e>0.25n_c$ when the laser amplitude is larger than a certain threshold. To satisfy the phase-matching conditions of frequency and wavenumber, the excited electrostatic mode has a constant frequency around half of the incident light frequency $\omega_0/2$, which is no longer the non-degenerate eigenmode of electron plasma wave $\omega_{pe}$. Both the scattered light and the electrostatic wave are trapped in plasma with their group velocities being zero. Super hot electrons are produced by the degenerate electrostatic wave. Our theoretical model is validated by particle-in-cell simulations. The SRS driven in this degenerate regime is an important laser energy loss mechanism in the laser plasma interactions as long as the laser intensity is higher than $10^{15}$W/cm$^2$.
Peer Review Status:Awaiting Review
Hits
Hits
Downloads
Comment