文件列表:

LICENSE (11357, 2022-11-07)
data/ (0, 2022-11-07)
data/data_prep.sh (545, 2022-11-07)
data/imus/ (0, 2022-11-07)
data/imus/00.mat (2075058, 2022-11-07)
data/imus/01.mat (500759, 2022-11-07)
data/imus/02.mat (2125057, 2022-11-07)
data/imus/04.mat (123250, 2022-11-07)
data/imus/05.mat (1259123, 2022-11-07)
data/imus/06.mat (501505, 2022-11-07)
data/imus/07.mat (501632, 2022-11-07)
data/imus/08.mat (1858474, 2022-11-07)
data/imus/09.mat (724373, 2022-11-07)
data/imus/10.mat (548396, 2022-11-07)
data_helper.py (1932, 2022-11-07)
evaluation.py (33119, 2022-11-07)
flops_target.zip (70143909, 2022-11-07)
helper.py (6310, 2022-11-07)
latency_target.zip (70660643, 2022-11-07)
layers.py (4197, 2022-11-07)
model.py (10928, 2022-11-07)
params.py (1692, 2022-11-07)
test.py (5849, 2022-11-07)
tools/ (0, 2022-11-07)
tools/__init__.py (0, 2022-11-07)
tools/pose_evaluation_utils.py (16877, 2022-11-07)
tools/transformations.py (66201, 2022-11-07)

# NASVIO This repository contains the evaluation code for *Search for efficient deep visual-inertial odometry through neural architecture search* and the searched models ## Data Preparation The test dataset is KITTI Odometry dataset. The IMU data after pre-processing is provided under `data/imus`. To download the images and poses, please run $cd data $source data_prep.sh ## Pretrained checkpoints on searched best models Two checkpoints with low FLOPS target (`flops_target.zip`) and low latency target (`latency_target.zip`) are provided. Simply unzip to retrieve the checkpoints. ## Test the pretrained models Select which model to run by changing the `self.target` parameter (`flops` or `latency`) in the `params.py`. Then run: python3 test.py

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：