Some questions about my dataset
supgy opened this issue · comments
Dear author,I want to use PTV3 for instance segmentation. My dataset is oral scan, which is its label indicates the category to which each point belongs and I organized them like scannet except no color. When I execute the training, the error occurs:
My config is as follws:
base = ["../base/default_runtime.py"]
misc custom setting
batch_size = 12 # bs: total bs in all gpus
num_worker = 24
mix_prob = 0.8
empty_cache = False
enable_amp = True
model settings
model = dict(
type="DefaultSegmentorV2",
#num_classes=20,
num_classes=17,
#backbone_out_channels=64,
backbone_out_channels=64,
backbone=dict(
type="PT-v3m1",
# in_channels=6,
in_channels=6,
order=("z", "z-trans", "hilbert", "hilbert-trans"),
stride=(2, 2, 2, 2),
enc_depths=(2, 2, 2, 6, 2),
enc_channels=(32, 64, 128, 256, 512),
enc_num_head=(2, 4, 8, 16, 32),
enc_patch_size=(1024, 1024, 1024, 1024, 1024),
dec_depths=(2, 2, 2, 2),
dec_channels=(64, 64, 128, 256),
dec_num_head=(4, 4, 8, 16),
dec_patch_size=(1024, 1024, 1024, 1024),
mlp_ratio=4,
qkv_bias=True,
qk_scale=None,
attn_drop=0.0,
proj_drop=0.0,
drop_path=0.3,
shuffle_orders=True,
pre_norm=True,
enable_rpe=False,
enable_flash=True,
upcast_attention=False,
upcast_softmax=False,
cls_mode=False,
pdnorm_bn=False,
pdnorm_ln=False,
pdnorm_decouple=True,
pdnorm_adaptive=False,
pdnorm_affine=True,
pdnorm_conditions=("ScanNet", "S3DIS", "Structured3D"),
),
criteria=[
dict(type="CrossEntropyLoss", loss_weight=1.0, ignore_index=-1),
dict(type="LovaszLoss", mode="multiclass", loss_weight=1.0, ignore_index=-1),
],
)
scheduler settings
epoch = 800
optimizer = dict(type="AdamW", lr=0.006, weight_decay=0.05)
scheduler = dict(
type="OneCycleLR",
max_lr=[0.006, 0.0006],
pct_start=0.05,
anneal_strategy="cos",
div_factor=10.0,
final_div_factor=1000.0,
)
param_dicts = [dict(keyword="block", lr=0.0006)]
dataset settings
dataset_type = "ScanNetDataset"
#data_root = "data/scannet"
data_root = '/opt/data/nvme2/gaoyang/tooth_scan/ptv3/'
data = dict(
#num_classes=20,
num_classes=17,
ignore_index=-1,
# names=["wall","floor", "cabinet","bed", "chair", "sofa", "table","door", "window", "bookshelf", "picture", "counter", "desk","curtain","refridgerator","shower curtain","toilet","sink","bathtub","otherfurniture",],
names=[
"ya1",
"ya2",
"ya3",
"ya4",
"ya5",
"ya6",
"ya7",
"ya8",
"ya9",
"ya10",
"ya11",
"ya12",
"ya13",
"ya14",
"ya15",
"ya16",
"yayin",
],
train=dict(
type=dataset_type,
split="train",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(
type="RandomDropout", dropout_ratio=0.2, dropout_application_ratio=0.2
),
# dict(type="RandomRotateTargetAngle", angle=(1/2, 1, 3/2), center=[0, 0, 0], axis="z", p=0.75),
dict(type="RandomRotate", angle=[-1, 1], axis="z", center=[0, 0, 0], p=0.5),
dict(type="RandomRotate", angle=[-1 / 64, 1 / 64], axis="x", p=0.5),
dict(type="RandomRotate", angle=[-1 / 64, 1 / 64], axis="y", p=0.5),
dict(type="RandomScale", scale=[0.9, 1.1]),
# dict(type="RandomShift", shift=[0.2, 0.2, 0.2]),
dict(type="RandomFlip", p=0.5),
dict(type="RandomJitter", sigma=0.005, clip=0.02),
dict(type="ElasticDistortion", distortion_params=[[0.2, 0.4], [0.8, 1.6]]),
dict(type="ChromaticAutoContrast", p=0.2, blend_factor=None),
dict(type="ChromaticTranslation", p=0.95, ratio=0.05),
dict(type="ChromaticJitter", p=0.95, std=0.05),
# dict(type="HueSaturationTranslation", hue_max=0.2, saturation_max=0.2),
# dict(type="RandomColorDrop", p=0.2, color_augment=0.0),
dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="train",
return_grid_coord=True,
),
dict(type="SphereCrop", point_max=102400, mode="random"),
dict(type="CenterShift", apply_z=False),
dict(type="NormalizeColor"),
# dict(type="ShufflePoint"),
dict(type="ToTensor"),
dict(
type="Collect",
keys=("coord", "grid_coord", "segment"),
#feat_keys=("color", "normal"),
feat_keys=("normal"),
),
],
test_mode=False,
),
val=dict(
type=dataset_type,
split="val",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="train",
return_grid_coord=True,
),
dict(type="CenterShift", apply_z=False),
dict(type="NormalizeColor"),
dict(type="ToTensor"),
dict(
type="Collect",
keys=("coord", "grid_coord", "segment"),
#feat_keys=("color", "normal"),
feat_keys=("normal"),
),
],
test_mode=False,
),
test=dict(
type=dataset_type,
split="val",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(type="NormalizeColor"),
],
test_mode=True,
test_cfg=dict(
voxelize=dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="test",
#keys=("coord", "color", "normal"),
keys=("coord", "normal"),
return_grid_coord=True,
),
crop=None,
post_transform=[
dict(type="CenterShift", apply_z=False),
dict(type="ToTensor"),
dict(
type="Collect",
keys=("coord", "grid_coord", "index"),
#feat_keys=("color", "normal"),
feat_keys=("normal"),
),
],
aug_transform=[
[
dict(
type="RandomRotateTargetAngle",
angle=[0],
axis="z",
center=[0, 0, 0],
p=1,
)
],
[
dict(
type="RandomRotateTargetAngle",
angle=[1 / 2],
axis="z",
center=[0, 0, 0],
p=1,
)
],
[
dict(
type="RandomRotateTargetAngle",
angle=[1],
axis="z",
center=[0, 0, 0],
p=1,
)
],
[
dict(
type="RandomRotateTargetAngle",
angle=[3 / 2],
axis="z",
center=[0, 0, 0],
p=1,
)
],
[
dict(
type="RandomRotateTargetAngle",
angle=[0],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[0.95, 0.95]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[1 / 2],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[0.95, 0.95]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[1],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[0.95, 0.95]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[3 / 2],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[0.95, 0.95]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[0],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[1.05, 1.05]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[1 / 2],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[1.05, 1.05]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[1],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[1.05, 1.05]),
],
[
dict(
type="RandomRotateTargetAngle",
angle=[3 / 2],
axis="z",
center=[0, 0, 0],
p=1,
),
dict(type="RandomScale", scale=[1.05, 1.05]),
],
[dict(type="RandomFlip", p=1)],
],
),
),
)
Can you teach me what's wrong?qaq Hope your reply.
Hi, I attached one of my configs for PTv3 + Instance Segmentation (experimental version), I forgot which is the best, but hope it can be a reference. I will check and release one next week.
Kindly remind:
- I think for PG-v1m2, I enabled seg criteria controlled with config, and I added Lovasz loss.
- Ignore the config and code name for PTv3, this is my experimental version.
Hope the demo is helpful!
_base_ = ["../_base_/default_runtime.py"]
# misc custom setting
batch_size = 12 # bs: total bs in all gpus
num_worker = 12
mix_prob = 0
empty_cache = False
enable_amp = True
evaluate = True
class_names = [
"wall",
"floor",
"cabinet",
"bed",
"chair",
"sofa",
"table",
"door",
"window",
"bookshelf",
"picture",
"counter",
"desk",
"curtain",
"refridgerator",
"shower curtain",
"toilet",
"sink",
"bathtub",
"otherfurniture",
]
num_classes = 20
segment_ignore_index = (-1, 0, 1)
# model settings
model = dict(
type="PG-v1m2",
backbone=dict(
type="PT-v5m1",
in_channels=6,
enc_depths=(2, 2, 2, 6, 2),
enc_channels=(32, 64, 128, 256, 512),
enc_num_head=(2, 4, 8, 16, 32),
enc_patch_size=(128, 128, 128, 128, 128),
dec_depths=(2, 2, 2, 2),
dec_channels=(64, 64, 128, 256),
dec_num_head=(4, 4, 8, 16),
dec_patch_size=(128, 128, 128, 128),
mlp_ratio=4,
qkv_bias=True,
qk_scale=None,
attn_drop=0.0,
proj_drop=0.0,
drop_path=0.3,
pre_norm=True,
order="z",
cls_mode=False,
),
backbone_out_channels=64,
semantic_num_classes=num_classes,
semantic_ignore_index=-1,
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
cluster_thresh=1.5,
cluster_closed_points=300,
cluster_propose_points=100,
cluster_min_points=50,
criteria=[
dict(type="CrossEntropyLoss", loss_weight=1.0, ignore_index=-1),
dict(
type="LovaszLoss",
mode="multiclass",
loss_weight=1.0,
ignore_index=-1,
),
],
)
# scheduler settings
epoch = 800
optimizer = dict(type="AdamW", lr=0.006, weight_decay=0.05)
scheduler = dict(
type="OneCycleLR",
max_lr=[0.006, 0.0006],
pct_start=0.05,
anneal_strategy="cos",
div_factor=10.0,
final_div_factor=1000.0,
)
param_dicts = [dict(keyword="block", lr=0.0006)]
# dataset settings
dataset_type = "ScanNetDataset"
data_root = "data/scannet"
data = dict(
num_classes=num_classes,
ignore_index=-1,
names=class_names,
train=dict(
type=dataset_type,
split="train",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(
type="RandomDropout", dropout_ratio=0.2, dropout_application_ratio=0.5
),
# dict(type="RandomRotateTargetAngle", angle=(1/2, 1, 3/2), center=[0, 0, 0], axis='z', p=0.75),
dict(type="RandomRotate", angle=[-1, 1], axis="z", center=[0, 0, 0], p=0.5),
dict(type="RandomRotate", angle=[-1 / 64, 1 / 64], axis="x", p=0.5),
dict(type="RandomRotate", angle=[-1 / 64, 1 / 64], axis="y", p=0.5),
dict(type="RandomScale", scale=[0.9, 1.1]),
# dict(type="RandomShift", shift=[0.2, 0.2, 0.2]),
dict(type="RandomFlip", p=0.5),
dict(type="RandomJitter", sigma=0.005, clip=0.02),
dict(type="ElasticDistortion", distortion_params=[[0.2, 0.4], [0.8, 1.6]]),
dict(type="ChromaticAutoContrast", p=0.2, blend_factor=None),
dict(type="ChromaticTranslation", p=0.95, ratio=0.1),
dict(type="ChromaticJitter", p=0.95, std=0.05),
# dict(type="HueSaturationTranslation", hue_max=0.2, saturation_max=0.2),
# dict(type="RandomColorDrop", p=0.2, color_augment=0.0),
dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="train",
return_grid_coord=True,
keys=("coord", "color", "normal", "segment", "instance"),
),
dict(type="SphereCrop", sample_rate=0.8, mode="random"),
dict(type="NormalizeColor"),
dict(
type="InstanceParser",
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
),
dict(type="ToTensor"),
dict(
type="Collect",
keys=(
"coord",
"grid_coord",
"segment",
"instance",
"instance_centroid",
"bbox",
),
feat_keys=("color", "normal"),
),
],
test_mode=False,
),
val=dict(
type=dataset_type,
split="val",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(
type="Copy",
keys_dict={
"coord": "origin_coord",
"segment": "origin_segment",
"instance": "origin_instance",
},
),
dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="train",
return_grid_coord=True,
keys=("coord", "color", "normal", "segment", "instance"),
),
# dict(type="SphereCrop", point_max=1000000, mode='center'),
dict(type="CenterShift", apply_z=False),
dict(type="NormalizeColor"),
dict(
type="InstanceParser",
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
),
dict(type="ToTensor"),
dict(
type="Collect",
keys=(
"coord",
"grid_coord",
"segment",
"instance",
"origin_coord",
"origin_segment",
"origin_instance",
"instance_centroid",
"bbox",
),
feat_keys=("color", "normal"),
offset_keys_dict=dict(offset="coord", origin_offset="origin_coord"),
),
],
test_mode=False,
),
test=dict(), # currently not available
)
hooks = [
dict(type="CheckpointLoader", keywords="module.", replacement="module."),
dict(type="IterationTimer", warmup_iter=2),
dict(type="InformationWriter"),
dict(
type="InsSegEvaluator",
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
),
dict(type="CheckpointSaver", save_freq=None),
]
Thank you very much for your reply! I try to use insseg-pointgroup-v1m1-0-spunet-base.py to perform instance segmentation,but it also have problem
Here is my config
_base_ = ["../_base_/default_runtime.py"]
# misc custom setting
batch_size = 12 # bs: total bs in all gpus
num_worker = 12
mix_prob = 0
empty_cache = False
enable_amp = True
evaluate = True
class_names = [
"ya1",
"ya2",
"ya3",
"ya4",
"ya5",
"ya6",
"ya7",
"ya8",
"ya9",
"ya10",
"ya11",
"ya12",
"ya13",
"ya14",
"ya15",
"ya16",
"yayin",
]
num_classes = 17
segment_ignore_index = (-1, 0, 1)
# model settings
model = dict(
type="PG-v1m1",
backbone=dict(
type="SpUNet-v1m1",
in_channels=6,
num_classes=17,
channels=(32, 64, 128, 256, 256, 128, 96, 96),
layers=(2, 3, 4, 6, 2, 2, 2, 2),
),
backbone_out_channels=64,
semantic_num_classes=num_classes,
semantic_ignore_index=-1,
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
cluster_thresh=1.5,
cluster_closed_points=300,
cluster_propose_points=100,
cluster_min_points=50,
)
# scheduler settings
epoch = 800
optimizer = dict(type="SGD", lr=0.1, momentum=0.9, weight_decay=0.0001, nesterov=True)
scheduler = dict(type="PolyLR")
# dataset settings
dataset_type = "ScanNetDataset"
data_root = '/opt/data/nvme2/gaoyang/tooth_scan/ptv3/'
data = dict(
num_classes=num_classes,
ignore_index=-1,
names=class_names,
train=dict(
type=dataset_type,
split="train",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(
type="RandomDropout", dropout_ratio=0.2, dropout_application_ratio=0.5
),
# dict(type="RandomRotateTargetAngle", angle=(1/2, 1, 3/2), center=[0, 0, 0], axis='z', p=0.75),
dict(type="RandomRotate", angle=[-1, 1], axis="z", center=[0, 0, 0], p=0.5),
dict(type="RandomRotate", angle=[-1 / 64, 1 / 64], axis="x", p=0.5),
dict(type="RandomRotate", angle=[-1 / 64, 1 / 64], axis="y", p=0.5),
dict(type="RandomScale", scale=[0.9, 1.1]),
# dict(type="RandomShift", shift=[0.2, 0.2, 0.2]),
dict(type="RandomFlip", p=0.5),
dict(type="RandomJitter", sigma=0.005, clip=0.02),
dict(type="ElasticDistortion", distortion_params=[[0.2, 0.4], [0.8, 1.6]]),
dict(type="ChromaticAutoContrast", p=0.2, blend_factor=None),
dict(type="ChromaticTranslation", p=0.95, ratio=0.1),
dict(type="ChromaticJitter", p=0.95, std=0.05),
# dict(type="HueSaturationTranslation", hue_max=0.2, saturation_max=0.2),
# dict(type="RandomColorDrop", p=0.2, color_augment=0.0),
dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="train",
return_grid_coord=True,
keys=("coord", "normal", "segment", "instance"),
),
dict(type="SphereCrop", sample_rate=0.8, mode="random"),
dict(type="NormalizeColor"),
dict(
type="InstanceParser",
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
),
dict(type="ToTensor"),
dict(
type="Collect",
keys=(
"coord",
"grid_coord",
"segment",
"instance",
"instance_centroid",
"bbox",
),
feat_keys=("normal",),
),
],
test_mode=False,
),
val=dict(
type=dataset_type,
split="val",
data_root=data_root,
transform=[
dict(type="CenterShift", apply_z=True),
dict(
type="Copy",
keys_dict={
"coord": "origin_coord",
"segment": "origin_segment",
"instance": "origin_instance",
},
),
dict(
type="GridSample",
grid_size=0.02,
hash_type="fnv",
mode="train",
return_grid_coord=True,
keys=("coord", "normal", "segment", "instance"),
),
# dict(type="SphereCrop", point_max=1000000, mode='center'),
dict(type="CenterShift", apply_z=False),
dict(type="NormalizeColor"),
dict(
type="InstanceParser",
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
),
dict(type="ToTensor"),
dict(
type="Collect",
keys=(
"coord",
"grid_coord",
"segment",
"instance",
"origin_coord",
"origin_segment",
"origin_instance",
"instance_centroid",
"bbox",
),
feat_keys=("normal",),
offset_keys_dict=dict(offset="coord", origin_offset="origin_coord"),
),
],
test_mode=False,
),
test=dict(), # currently not available
)
hooks = [
dict(type="CheckpointLoader", keywords="module.", replacement="module."),
dict(type="IterationTimer", warmup_iter=2),
dict(type="InformationWriter"),
dict(
type="InsSegEvaluator",
segment_ignore_index=segment_ignore_index,
instance_ignore_index=-1,
),
dict(type="CheckpointSaver", save_freq=None),
]