CoLC/train_colc.bash at main · CatOneTwo/CoLC · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

DATASET=v2xsim
SETTING=lidar_only_with_noise


# 1. Prepare early fusion model
########## Model prepare ##########

# CUDA_VISIBLE_DEVICES=0 python opencood/tools/train.py \
# -y opencood/hypes_yaml/${DATASET}/${SETTING}/pointpillar_early.yaml \
# --fusion_method early

########## Model prepare ##########


# 2. Pretrain point selector
########## point selector ##########

CUDA_VISIBLE_DEVICES=0 python opencood/tools/train.py \
-y opencood/hypes_yaml/${DATASET}/${SETTING}/pointnet_single.yaml \
--run_test False \
--log early

########## point selector ##########


# 3. Pretrain vq-based completion module
########## lidar completion ##########

# CUDA_VISIBLE_DEVICES=0 python opencood/tools/train_vqvae_ga_new.py \
# -y opencood/hypes_yaml/${DATASET}/${SETTING}/pointpillar_colc_baseline_vqvae.yaml \
# --pretrained_model opencood/logs/${DATASET}_point_pillar_lidar_early  # early fusion model

########## lidar completion ##########


# 4. Train CoLC model
########## CoLC ##########
# CUDA_VISIBLE_DEVICES=0 python opencood/tools/train_colc_ga.py \
# -y opencood/hypes_yaml/${DATASET}/${SETTING}/pointpillar_colc_kd.yaml \
# --fusion_method cecooper \
# --nei_model opencood/logs/${DATASET}_pointnet_lidar_seg_early \
# --pretrained_model opencood/logs/${DATASET}_point_pillar_lidar_early \
# --vqvae_model opencood/logs/$completion_model_path$
########## CoLC ##########