Search for: All records

A major challenge in monocular 3D object detection is the limited diversity and quantity of objects in real datasets. While augmenting real scenes with virtual objects holds promise to improve both the diversity and quantity of the objects, it remains elusive due to the lack of an effective 3D object insertion method in complex real captured scenes. In this work, we study augmenting complex real indoor scenes with virtual objects for monocular 3D object detection. The main challenge is to automatically identify plausible physical properties for virtual assets (e.g., locations, appearances, sizes, etc.) in cluttered real scenes. To address this challenge, we propose a physically plausible indoor 3D object insertion approach to automatically copy virtual objects and paste them into real scenes. The resulting objects in scenes have 3D bounding boxes with plausible physical locations and appearances. In particular, our method first identifies physically feasible locations and poses for the inserted objects to prevent collisions with the existing room layout. Subsequently, it estimates spatially-varying illumination for the insertion location, enabling the immersive blending of the virtual objects into the original scene with plausible appearances and cast shadows. We show that our augmentation method significantly improves existing monocular 3D object models and achieves state-of-the-art performance. For the first time, we demonstrate that a physically plausible 3D object insertion, serving as a generative data augmentation technique, can lead to significant improvements for discriminative downstream tasks such as monocular 3D object detection. 

Hierarchically porous electrodes made of electrochemically active materials and conductive additives may display synergistic effects originating from the interactions between the constituent phases, and this approach has been adopted for optimizing the performances of many electrode materials. Here we report our findings in design, fabrication, and characterization of a hierarchically porous hybrid electrode composed of α-NiS nanorods decorated on reduced graphene oxide (rGO) (denoted as R-NiS/rGO), derived from water-refluxed metal–organic frameworks/rGO (Ni-MOF-74/rGO) templates. Microanalyses reveal that the as-synthesized α-NiS nanorods have abundant (101) and (110) surfaces on the edges, which exhibit a strong affinity for OH − in KOH electrolyte, as confirmed by density functional theory-based calculations. The results suggest that the MOF-derived α-NiS nanorods with highly exposed active surfaces are favorable for fast redox reactions in a basic electrolyte. Besides, the presence of rGO in the hybrid electrode greatly enhances the electronic conductivity, providing efficient current collection for fast energy storage. Indeed, when tested in a supercapacitor with a three-electrode configuration in 2 M KOH electrolyte, the R-NiS/rGO hybrid electrode exhibits a capacity of 744 C g −1 at 1 A g −1 and 600 C g −1 at 50 A g −1 , indicating remarkable rate performance, while maintaining more than 89% of the initial capacity after 20 000 cycles. Moreover, when coupled with a nitrogen-doped graphene aerogel (C/NG-A) negative electrode, the hybrid supercapacitor (R-NiS/rGO/electrolyte/C/NG-A) achieved an ultra-high energy density of 93 W h kg −1 at a power density of 962 W kg −1 , while still retaining an energy density of 54 W h kg −1 at an elevated working power of 46 034 W kg −1 . 

Abstract Graphene has a great potential to replace silicon in prospective semiconductor industries due to its outstanding electronic and transport properties; nonetheless, its lack of energy bandgap is a substantial limitation for practical applications. To date, straining graphene to break its lattice symmetry is perhaps the most efficient approach toward realizing bandgap tunability in graphene. However, due to the weak lattice deformation induced by uniaxial or in‐plane shear strain, most strained graphene studies have yielded bandgaps <1 eV. In this work, a modulated inhomogeneous local asymmetric elastic–plastic straining is reported that utilizes GPa‐level laser shocking at a high strain rate (dε/dt) ≈ 10⁶–10⁷s⁻¹, with excellent formability, inducing tunable bandgaps in graphene of up to 2.1 eV, as determined by scanning tunneling spectroscopy. High‐resolution imaging and Raman spectroscopy reveal strain‐induced modifications to the atomic and electronic structure in graphene and first‐principles simulations predict the measured bandgap openings. Laser shock modulation of semimetallic graphene to a semiconducting material with controllable bandgap has the potential to benefit the electronic and optoelectronic industries.