search

CGuangyan-BIT / PointGPT

[NeurIPS 2023] PointGPT: Auto-regressively Generative Pre-training from Point Clouds
MIT License
179 stars 18 forks source link

Visualization of the result #16

Closed colin-de closed 7 months ago

colin-de commented 8 months ago

Hi, where can we get the visualization of the result?

CGuangyan-BIT commented 8 months ago

Hi! You can use the main_vis.py file for visualization (please see part 6 of the readme), and the results will be saved in the ./vis directory

colin-de commented 8 months ago

Hi! thanks for the quick reply. i used this the command --test --ckpts data/ModelNet/modelnet_1k.pth --config cfgs/PointGPT-S/finetune_modelnet.yaml --exp_name finetune and got

/home/wli/miniconda3/envs/pointgpt/bin/python /home/wli/code/PointGPT/main_vis.py --test --ckpts data/ModelNet/modelnet_1k.pth --config cfgs/PointGPT-S/finetune_modelnet.yaml --exp_name finetune 
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - Copy the Config file from cfgs/PointGPT-S/finetune_modelnet.yaml to ./experiments/finetune_modelnet/PointGPT-S/test_finetune/config.yaml
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.config : cfgs/PointGPT-S/finetune_modelnet.yaml
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.launcher : none
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.local_rank : 0
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.num_workers : 8
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.seed : 0
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.deterministic : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.sync_bn : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.exp_name : test_finetune
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.loss : cd1
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.start_ckpts : None
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.ckpts : data/ModelNet/modelnet_1k.pth
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.val_freq : 1
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.vote : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.resume : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.test : True
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.finetune_model : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.scratch_model : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.mode : None
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.way : -1
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.shot : -1
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.fold : -1
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.experiment_path : ./experiments/finetune_modelnet/PointGPT-S/test_finetune
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.tfboard_path : ./experiments/finetune_modelnet/PointGPT-S/TFBoard/test_finetune
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.log_name : finetune_modelnet
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.use_gpu : True
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - args.distributed : False
2023-11-17 16:53:31,451 - finetune_modelnet - INFO - config.optimizer = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.optimizer.type : AdamW
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.optimizer.kwargs = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.optimizer.kwargs.lr : 0.0001
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.optimizer.kwargs.weight_decay : 0.05
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.scheduler = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.scheduler.type : CosLR
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.scheduler.kwargs = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.scheduler.kwargs.epochs : 300
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.scheduler.kwargs.initial_epochs : 10
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train._base_ = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train._base_.NAME : ModelNet
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train._base_.DATA_PATH : data/ModelNet/modelnet40_normal_resampled
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train._base_.N_POINTS : 8192
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train._base_.NUM_CATEGORY : 40
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train._base_.USE_NORMALS : False
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train.others = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train.others.subset : train
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.train.others.bs : 128
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val._base_ = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val._base_.NAME : ModelNet
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val._base_.DATA_PATH : data/ModelNet/modelnet40_normal_resampled
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val._base_.N_POINTS : 8192
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val._base_.NUM_CATEGORY : 40
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val._base_.USE_NORMALS : False
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val.others = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val.others.subset : test
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.val.others.bs : 1
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test._base_ = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test._base_.NAME : ModelNet
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test._base_.DATA_PATH : data/ModelNet/modelnet40_normal_resampled
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test._base_.N_POINTS : 8192
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test._base_.NUM_CATEGORY : 40
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test._base_.USE_NORMALS : False
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test.others = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test.others.subset : test
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.dataset.test.others.bs : 1
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model = edict()
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.NAME : PointTransformer
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.trans_dim : 384
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.depth : 12
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.drop_path_rate : 0.1
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.cls_dim : 40
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.num_heads : 6
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.group_size : 32
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.num_group : 64
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.encoder_dims : 384
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.decoder_depth : 4
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.loss : cdl2
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.model.weight_center : 1
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.npoints : 1024
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.total_bs : 128
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.step_per_update : 1
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.max_epoch : 300
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - config.grad_norm_clip : 10
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - Distributed training: False
2023-11-17 16:53:31,452 - finetune_modelnet - INFO - Set random seed to 0, deterministic: False
2023-11-17 16:53:31,453 - finetune_modelnet - INFO - Tester start ... 
2023-11-17 16:53:31,455 - ModelNet - INFO - The size of test data is 2468
2023-11-17 16:53:31,455 - ModelNet - INFO - Load processed data from data/ModelNet/modelnet40_normal_resampled/modelnet40_test_8192pts_fps.dat...
2023-11-17 16:53:31,792 - finetune_modelnet - INFO - Loading weights from data/ModelNet/modelnet_1k.pth...
2023-11-17 16:53:31,986 - finetune_modelnet - INFO - ckpts @ 185 epoch( performance = {'acc': tensor(93.2739)})

Process finished with exit code 0

and the visualization is not saved and exited

CGuangyan-BIT commented 8 months ago

Hi! This is because main_vis.py is for the visualization of the pretrain generation task, and you may need to change the config to pretrain.yaml

colin-de commented 7 months ago

Thanks！