-
Notifications
You must be signed in to change notification settings - Fork 319
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Distributed training [SemSeg] #530
base: main
Are you sure you want to change the base?
Conversation
This pull request introduces 45 alerts and fixes 6 when merging 4ff5fbe into c461790 - view on LGTM.com new alerts:
fixed alerts:
|
This pull request introduces 45 alerts and fixes 6 when merging 1803d77 into c461790 - view on LGTM.com new alerts:
fixed alerts:
|
This pull request introduces 45 alerts and fixes 6 when merging c8f4589 into c461790 - view on LGTM.com new alerts:
fixed alerts:
|
This pull request introduces 44 alerts when merging 40f888f into 1c45bfe - view on LGTM.com new alerts:
|
This pull request introduces 33 alerts and fixes 1 when merging 375bb8f into 1c45bfe - view on LGTM.com new alerts:
fixed alerts:
|
This pull request introduces 33 alerts and fixes 1 when merging 50c8c20 into 1c45bfe - view on LGTM.com new alerts:
fixed alerts:
|
This pull request introduces 8 alerts and fixes 4 when merging 5c2ba32 into 1c45bfe - view on LGTM.com new alerts:
fixed alerts:
|
This pull request introduces 8 alerts and fixes 4 when merging e971747 into 1c45bfe - view on LGTM.com new alerts:
fixed alerts:
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewed 18 of 23 files at r1.
Reviewable status: 18 of 23 files reviewed, 4 unresolved discussions (waiting on @sanskar107 and @ssheorey)
ml3d/configs/sparseconvunet_scannet.yml
line 9 at r1 (raw file):
use_cache: False sampler: name: None
Why change from 'SemsegRandomSampler' to None?
ml3d/configs/default_cfgs/kitti360.yml
line 2 at r1 (raw file):
name: KITTI360 dataset_path: /Users/sanskara/data/kitti360/
hardcoded path
ml3d/torch/models/sparseconvnet.py
line 119 at r1 (raw file):
feat = np.array(data['feat'], dtype=np.float32) if feat.shape[1] < 3: feat = np.concatenate([feat, np.ones([feat.shape[0], 1])], 1)
Why force the feat size to be >= 3?
ml3d/torch/modules/metrics/semseg_metric.py
line 30 at r1 (raw file):
def __iadd__(self, otherMetric): if self.confusion_matrix is None and otherMetric.confusion_matrix is None: pass
Why not throw an exception?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewable status: 18 of 23 files reviewed, 22 unresolved discussions (waiting on @benjaminum, @sanskar107, and @ssheorey)
ml3d/datasets/kitti360.py
line 68 at r1 (raw file):
raise ValueError( "Invalid Path, make sure dataset_path is the parent directory of data_3d_semantics." )
[nit] catch error from open and throw informative message instead of a separate check.
ml3d/datasets/kitti360.py
line 160 at r1 (raw file):
attr: The attributes that correspond to the outputs passed in results. """
move docstring after function signature.
ml3d/datasets/nuscenes_semseg.py
line 63 at r1 (raw file):
self.val_info = {} if os.path.exists(join(info_path, 'infos_train.pkl')):
open() will throw an error if path is not readable - no need to separately check exists().
Also for the 2 following files.
ml3d/datasets/nuscenes_semseg.py
line 111 at r1 (raw file):
} self.label_mapping = np.array( [mapping[i] for i in range(0, len(mapping))], dtype=np.int32)
np.fromiter(mapping.values(), dtype=np.int32)
should be the same
ml3d/datasets/nuscenes_semseg.py
line 122 at r1 (raw file):
""" classes = "ignore, barrier, bicycle, bus, car, construction_vehicle, motorcycle, pedestrian, traffic_cone, trailer, trucl, driveable_surface, other_flat, sidewalk, terrain, manmade, vegetation" classes = classes.replace(', ', ',').split(',')
can directly split(', ')
ml3d/datasets/nuscenes_semseg.py
line 136 at r1 (raw file):
A data object with lidar information. """ assert Path(path).exists()
assert not required.
ml3d/datasets/nuscenes_semseg.py
line 146 at r1 (raw file):
A data object with semantic information. """ assert Path(path).exists()
assert not required
ml3d/datasets/nuscenes_semseg.py
line 199 at r1 (raw file):
attribute is stored; else, returns false. """ pass
raise NotImplementedError() instead of pass to prevent silent bugs
ml3d/datasets/nuscenes_semseg.py
line 210 at r1 (raw file):
results. """ pass
raise NotImplementedError() instead of pass to prevent silent bugs
ml3d/datasets/waymo_semseg.py
line 81 at r1 (raw file):
""" classes = "Undefined, Car, Truck, Bus, Other Vehicle, Motorcyclist, Bicyclist, Pedestrian, Sign, Traffic Light, Pole, Construction Cone, Bicycle, Motorcycle, Building, Vegetation, Tree Trunk, Curb, Road, Lane Marker, Other Ground, Walkable, Sidewalk" classes = classes.replace(', ', ',').split(',')
can directly split(", ")
without replace()
ml3d/torch/pipelines/semantic_segmentation.py
line 164 at r1 (raw file):
for unused_step, inputs in enumerate(infer_loader): if hasattr(inputs['data'], 'to'): inputs['data'].to(device)
Isn't inputs['data']
always a torch tensor, so it should always have ato()
function?
ml3d/torch/pipelines/semantic_segmentation.py
line 364 at r1 (raw file):
if train_sampler is not None or valid_sampler is not None: raise NotImplementedError( "Distributed training with sampler is not supported yet!")
Collect any limitations and future TODOs such as this in the PR description.
scripts/preprocess_waymo.py
line 69 at r1 (raw file):
def __init__(self, dataset_path, save_dir='', workers=8, is_test=False): self.write_image = True
Remove if from debugging
scripts/preprocess_waymo.py
line 138 at r1 (raw file):
frame.context.stats.location not in self.selected_waymo_locations): print("continue")
Is this from debugging? Remove if so.
scripts/preprocess_waymo.py
line 211 at r1 (raw file):
'w+') as fp_calib: fp_calib.write(calib_context) fp_calib.close()
close()
is redundant with context manager open()
scripts/preprocess_waymo.py
line 260 at r1 (raw file):
# if self.filter_empty_3dboxes and obj.num_lidar_points_in_box < 1: # continue
remove commented code.
scripts/preprocess_waymo.py
line 272 at r1 (raw file):
# pt_ref = self.T_velo_to_front_cam @ \ # np.array([x, y, z, 1]).reshape((4, 1)) # x, y, z, _ = pt_ref.flatten().tolist()
remove commented code
scripts/preprocess_waymo_semseg.py
line 41 at r1 (raw file):
parser.add_argument( '--out_path', help='Output path to store pickle (default to dataet_path)',
typo
This change is