-
-
- https://arxiv.org/abs/2103.07351
- 2021
信頼性の高い正確な3Dトラッキングフレームワークは、自律走行などの多くのアプリケーションにおいて、周囲の物体の将来の位置を予測し、観察者の行動を計画するために不可欠である。
本研究では、移動するプラットフォーム上で撮影された一連の2D画像から、時間の経過とともに移動するオブジェクトを効果的に関連付け、その…
e4exp updated
3 years ago
-
Hello @mrnabati. Wishing you a pleasant day. I would like to know the difference in the 2 pre-trained models that you mentioned in the readme file. The one is the centernet baseline model with 170 epo…
-
mask = mask[::-1, ::-1]
在看源码时，发现这里对mask进行了180度的选择，请问为什么要这么设计呢？
-
Thanks for your amazing work in enhancing the performance of the downstream AV tasks. Could you please elaborate on the process you used to generate your gen-nuScenes dataset? Additionally, what were …
-
Issue #1002 seems to indicate that multisweeps is NOT used on the Waymo dataset models.
Yet in the preprocessing code for Waymo, there is a `max_sweeps` param seen [here](https://github.com/open-mm…
-
I am attempting to train the UNET but it doesn't seem to be learning and just spits out garbage values. I even tried to overfit it to a single example but it doesn't even learn that. Ill attach a pdf …
-
hillo，I trained the pandaset dataset with your code。The training went well with transfusion_pillar_L and training log see
link：https://pan.baidu.com/s/1PFjYIedIrYY1qKw5SwUe1A?pwd=4cho
Extraction cod…
-
我用与原文相同的配置训练：
cmt_voxel0100_r50_800x320_cbgs.py
bs: samples_per_gpu=2, 8卡A100
lr：0.00014
target_ratio: (6, …
-
hello! thanks for your great works!
when i go into source code, i found "sweep", "ida" and "bda". these maybe something we have not seen in LSS. can you kindly explain the meaning of the three conc…