Skip to content

Latest commit

 

History

History
59 lines (49 loc) · 2.77 KB

README.md

File metadata and controls

59 lines (49 loc) · 2.77 KB

Data Preparation

Download Scene Flow Datasets, KITTI 2012, KITTI 2015

Scene Flow Datasets

Please update DATAPATH in the sceneflow.sh file as your training data path.

Training

Multi-GPU training

export NGPUS=8
export save_results_dir="./checkpoints"
./scripts/sceneflow.sh $NGPUS $save_results_dir  --ndisps "48,24"  --disp_inter_r "4,1"   --dlossw "0.5,1.0,2.0"  --batch_size 2 --eval_freq 3  --model gwcnet-c

You can set --mdoel gwcnet to using GwcNet-gc as baseline model. gwcnet-c refer to only using concatenation cost volume which similar to PSMNet.

Evaluation

export NGPUS=8
export save_results_dir="./checkpoints"
./scripts/sceneflow.sh $NGPUS $save_results_dir --loadckpt $CKPT_FILE- --ndisps "48,24"  --disp_inter_r "4,1"   --batch_size 2 --mode test  --model gwcnet-c

Results on SceneFlow datsets

>1px. >2px. >3px. EPE.
GwcNet-gc 8.03 4.47 3.30 0.765
CasGwcNet-gc(D=48,24) 7.47 4.16 3.05 0.649
GwcNet-c 8.41 4.63 3.41 0.808
CasGwcNet-c(D=48,24) 7.49 4.16 3.04 0.650

KITTI 2012 / 2015

Please update DATAPATH in the kitti15.sh file as your training data path.

Training

export NGPUS=8
export save_results_dir="./checkpoints"
./scripts/kitti15.sh $NGPUS $save_results_dir  --ndisps "48,24"  --disp_inter_r "4,1"   --dlossw "0.5,1.0,2.0"  --batch_size 2 --eval_freq 3  --model gwcnet-c

Evaluation

  • Set CKPT_FILE as your checkpoint file, you also can download my pretrained model cas-gwcnet-c for KITTI 2015.
export NGPUS=8
export save_results_dir="./checkpoints"
./scripts/kitti15.sh $NGPUS $save_results_dir  --loadckpt $CKPT_FILE- --ndisps "48,24"  --disp_inter_r "4,1"   --batch_size 2 --mode test  --model gwcnet-c

Save Disps

export save_path="./outputs/predictions"
./scripts/kitti15_save.sh $save_results_dir  --loadckpt $CKPT_FILE --ndisps "48,24"  --disp_inter_r "4,1"   --batch_size 2 --model gwcnet-c