Stereo Neural Vernier Caliper

Authors: Shichao Li, Zechun Liu, Zhiqiang Shen, Kwang-Ting Cheng1376-1385

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments demonstrate the effectiveness of our approach, which achieves state-of-the-art performance on the KITTI benchmark.
Researcher Affiliation Academia 1Hong Kong University of Science and Technology 2Carnegie Mellon University 3Mohamed bin Zayed University of Artificial Intelligence
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Code and pre-trained models are available at https://github.com/Nicholasli1995/SNVC.
Open Datasets Yes We employ the KITTI object detection benchmark (Geiger, Lenz, and Urtasun 2012) for evaluation, which contains outdoor RGB images captured with calibrated stereo cameras.
Dataset Splits Yes The dataset is split into 7,481 training images and 7,518 testing images. The training images are further split into the train split and the val split containing 3,712 and 3,769 images respectively.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments. It mentions training, but no hardware specifics.
Software Dependencies No The paper does not provide specific version numbers for software dependencies or libraries used in the experiments. It mentions HRNet-W32 but not its specific version or other software.
Experiment Setup Yes In experiments the standard derivations for the above attributes are 0.3m, 0m, 0.3m, 5cm, 5cm, 5cm and 5 respectively. We use the default parameters γ = 2 and α = 0.25 and the total training loss LA total = Lconf + Lcoord + Lfg. In experiments, we specify L, H, W as 3, 10, and 3 centimeters respectively and [NL, NH, NW ] = [192, 32, 128].