CasMVSNet+Ours

CasMVSNet+DFPN

Training

Please see each subsection for training on different datasets. Available training datasets:

• DTU dataset
• BlendedMVS

DTU dataset

Data download

Download the preprocessed DTU training data and Depth_raw from original MVSNet repo and unzip. For the description of how the data is created, please refer to the original paper.

Training model

Run (example)

python train.py \
   --dataset_name dtu \
   --root_dir $DTU_DIR \
   --num_epochs 16 --batch_size 2 \
   --depth_interval 2.65 --n_depths 8 32 48 --interval_ratios 1.0 2.0 4.0 \
   --optimizer adam --lr 1e-3 --lr_scheduler cosine \
   --exp_name exp

Note that the model consumes huge GPU memory, so the batch size is generally small.

See opt.py for all configurations.

---

BlendedMVS

Run

python train.py \
   --dataset_name blendedmvs \
   --root_dir $BLENDEDMVS_LOW_RES_DIR \
   --num_epochs 16 --batch_size 2 \
   --depth_interval 192.0 --n_depths 8 32 48 --interval_ratios 1.0 2.0 4.0 \
   --optimizer adam --lr 1e-3 --lr_scheduler cosine \
   --exp_name exp

The --depth_interval 192.0 is the product of the coarsest n_depth and the coarsest --interval_ratio: 192.0=48x4.0.

Testing

Dtu

Run

python eval.py --dataset_name dtu \
  --root_dir MVS/dtu_test \
  --img_wh 1152 864 \
  --ckpt_path epoch=15.ckpt \
  --deform_conv 0 1 0 0 1 0 0 1 \
  --split test \
  --conf 0.1 \
  --min_geo_consistent 5 \
  #--save_visual \
  #--sacn $SCAN

Blended

Run

python eval.py --dataset_name blendedmvs \
  --root_dir dataset_low_res \
  --img_wh 768 576 \
  --ckpt_path epoch=15.ckpt \
  --save_visual \
  --deform_conv 0 1 0 0 1 0 0 1 \
  --split val \
  --conf 0.1 \
  --min_geo_consistent 5 \
  --depth_interval 192 \
  #--save_visual \
  #--sacn $SCAN

Tanks and Temples

Run

python eval.py --dataset_name tanks \ 	--root_dir tankandtemples/ \
  --ckpt_path epoch=15.ckpt \
  --deform_conv 0 1 0 0 1 0 0 1 \
  --split intermediate \
  --min_geo_consistent 5 \
  #--save_visual \
  #--sacn $SCAN