1 code implementation • 18 Apr 2024 • Yubo Gao, Maryam Haghifam, Christina Giannoula, Renbo Tu, Gennady Pekhimenko, Nandita Vijaykumar
Development of new DL models typically involves two parties: the model developers and performance optimizers.
no code implementations • 1 Mar 2024 • Chunlin Li, Ruofan Liang, Hanrui Fan, Zhengen Zhang, Sankeerth Durvasula, Nandita Vijaykumar
We present a framework, DISORF, to enable online 3D reconstruction and visualization of scenes captured by resource-constrained mobile robots and edge devices.
no code implementations • 1 Dec 2023 • Sankeerth Durvasula, Adrian Zhao, Fan Chen, Ruofan Liang, Pawan Kumar Sanjaya, Nandita Vijaykumar
In this work, we observe that the gradient computation phase during training is a significant bottleneck on GPUs due to the large number of atomic operations that need to be processed.
1 code implementation • ICCV 2023 • Ruofan Liang, Huiting Chen, Chunlin Li, Fan Chen, Selvakumar Panneer, Nandita Vijaykumar
In this work, we introduce ENVIDR, a rendering and modeling framework for high-quality rendering and reconstruction of surfaces with challenging specular reflections.
no code implementations • 8 Mar 2023 • Sankeerth Durvasula, Yushi Guan, Nandita Vijaykumar
We also demonstrate a speedup of up to 1. 6X for inference using CNNs for tasks such as depth estimation, object recognition, and optical flow estimation, with almost no loss in accuracy.
no code implementations • 15 Oct 2022 • Ruofan Liang, Jiahao Zhang, Haoda Li, Chen Yang, Yushi Guan, Nandita Vijaykumar
To enable more accurate illumination updates after deformation, we use the shadow mapping technique to approximate the light visibility updates caused by geometry editing.
no code implementations • ICLR 2022 • Ruofan Liang, Hongyi Sun, Nandita Vijaykumar
In this work, we aim to accelerate inference and training of coordinate-based MLPs for implicit neural representations by proposing a new split MLP architecture, CoordX.
no code implementations • 22 May 2018 • Bojian Zheng, Abhishek Tiwari, Nandita Vijaykumar, Gennady Pekhimenko
For each feature map recomputation to be effective and efficient, its effect on (1) the total memory footprint, and (2) the total execution time has to be carefully estimated.