search

PointsCoder / GPT-Driver

Learning to Drive with GPT
https://pointscoder.github.io/projects/gpt_driver/index.html
224 stars 11 forks source link

Cached_nuscenes_info.pkl file creation #3

Open abhigoku10 opened 10 months ago

abhigoku10 commented 10 months ago

@PointsCoder Thanks for sharing the code base, can u please share the script to create the cached_nuscenes_info.pkl files

THanks in advance

PointsCoder commented 7 months ago

@abhigoku10 I refer mostly to the cache code in VAD but it is on another codebase. I currently don't have plans to immigrate to this repo.

abhigoku10 commented 7 months ago

@PointsCoder Thanks for the response can u please share the link of VAD which is used to create the cached_nuscenes_info.pkl file

cherry956 commented 7 months ago

@PointsCoder I see your code, but I don't understand how to create the cached_nuscenes_info.pkl files, because I need to add some information adout vehicle. can u share the code with me , thanks.

abhigoku10 commented 7 months ago

@PointsCoder Please find my updated queries

  1. I have followed this link https://github.com/hustvl/VAD to create the data "cached_nuscenes_info.pkl" , but when the file is created there are missing keys in the data ie 'gt_ego_his_diff', 'gt_ego_fut_diff' . Have u added it additionally ??

2.From UNIAD when I create the "detection_motion_result_trainval.json" there is a difference in the content of the data ie shown below attn_mask:None

command:array([2])

is_sdc:True

nusc_box:label: 0, score: 0.39, xyz: [0.04, -0.04, 0.00], wlh: [1.73, 4.07, 1.55], rot axis: [-0.00, -0.00, -1.00], ang(degrees): -90.01, ang(rad): -1.57, vel: nan, nan, nan, name: None, token: None

past_pred_traj:None

pred_center:[0, 0, -0.41999999999999993]

pred_dim:array([1.7324345, 4.0702405, 1.5549785], dtype=float32)

pred_label:0

pred_occ_map:None

pred_score:0.39475283

pred_track_id:-1

pred_traj:array([[ 0.06253336, 3.4623034 ], [ 0.15860406, 6.559455 ], [ 0.2630881 , 9.370248 ], [ 0.36474904, 11.905737 ], [ 0.4004062 , 14.156303 ], [ 0.30280203, 16.245867 ]], dtype=float32)

pred_traj_max:array([[ 0.06253336, 3.4623034 ], [ 0.15860406, 6.559455 ], [ 0.2630881 , 9.370248 ], [ 0.36474904, 11.905737 ], [ 0.4004062 , 14.156303 ], [ 0.30280203, 16.245867 ]], dtype=float32)

pred_traj_score:1

pred_vel:array([0.04977103, 6.916191 ], dtype=float32)

pred_yaw:-4.7122004787114005 has context menu

is this correct ?? How did u generate the file

PointsCoder commented 7 months ago

@abhigoku10 @cherry956 nuScenes data caching should be here if I recalled correctly.

For customized caching, please play with it yourself.