[Question]What is the "post-processing"

[GarryJAY502]

❓ Question

When I running the preprocess.py,there is a warning
2024-11-24 21:52:01.503 | WARNING | nndet.planning.experiment.base:determine_postprocessing:263 - No planning for post-processing implemented.

2024-11-24 21:51:28.019 | INFO | nndet.planning.properties.intensity:run_collect_intensity_properties:86 - Processing intensity values of modality 0 2024-11-24 21:52:01.497 | INFO | nndet.planning.experiment.base:determine_whether_to_use_mask_for_norm:311 - not using nonzero mask for normalization 2024-11-24 21:52:01.497 | INFO | nndet.planning.experiment.base:plan_base:164 - Are we using the nonzero maks for normalization? OrderedDict([(0, False)]) 2024-11-24 21:52:01.500 | INFO | nndet.planning.experiment.base:plan_base:166 - Base target spacing is [5. 0.42969301 0.42969301] 2024-11-24 21:52:01.503 | INFO | nndet.planning.experiment.base:plan_base:169 - Normalization schemes OrderedDict([(0, 'nonCT')]) **2024-11-24 21:52:01.503 | WARNING | nndet.planning.experiment.base:determine_postprocessing:263 - No planning for post-processing implemented.** 2024-11-24 21:52:01.505 | INFO | nndet.planning.experiment.base:plan_base_stage:219 - The median shape of the dataset is [ 35.6 511.99344824 511.99165494] 2024-11-24 21:52:01.506 | INFO | nndet.planning.experiment.base:plan_base_stage:221 - The max shape in the dataset is [ 51.48000069 605.08312736 605.0873354 ] 2024-11-24 21:52:01.506 | INFO | nndet.planning.experiment.base:plan_base_stage:223 - The min shape in the dataset is [ 16. 464.2364498 388.99727502] 2024-11-24 21:52:01.507 | INFO | nndet.planning.experiment.base:plan_base_stage:227 - The transposed median shape of the dataset is [ 35.6 511.99344824 511.99165494] 2024-11-24 21:52:01.529 | INFO | nndet.planning.architecture.boxes.c002:process_properties:73 - Processing dataset properties 2024-11-24 21:52:04.636 | INFO | nndet.planning.architecture.boxes.c002:_get_initial_patch_size:340 - Using initial patch size: [ 36 512 512] 2024-11-24 21:52:04.636 | INFO | nndet.ptmodule.retinaunet.base:from_config_plan:362 - Architecture overwrites: {} Anchor overwrites: {} 2024-11-24 21:52:04.637 | INFO | nndet.ptmodule.retinaunet.base:from_config_plan:364 - Building architecture according to plan of not_found 2024-11-24 21:52:04.637 | INFO | nndet.ptmodule.retinaunet.base:from_config_plan:367 - Start channels: 32; head channels: 128; fpn channels: 128 2024-11-24 21:52:04.637 | INFO | nndet.core.boxes.anchors:__init__:288 - Discarding anchor generator kwargs {'stride': 1} 2024-11-24 21:52:04.637 | INFO | nndet.ptmodule.retinaunet.base:_build_encoder:464 - Building:: encoder Encoder: {} 2024-11-24 21:52:04.817 | INFO | nndet.ptmodule.retinaunet.base:_build_decoder:496 - Building:: decoder UFPNModular: {'min_out_channels': 8, 'upsampling_mode': 'transpose', 'num_lateral': 1, 'norm_lateral': False, 'activation_lateral': False, 'num_out': 1, 'norm_out': False, 'activation_out': False} 2024-11-24 21:52:04.844 | INFO | nndet.core.boxes.matcher.atss:__init__:45 - Running ATSS Matching with num_candidates=4 and center_in_gt False. 2024-11-24 21:52:04.845 | INFO | nndet.ptmodule.retinaunet.base:_build_head_classifier:530 - Building:: classifier BCECLassifier: {'num_convs': 1, 'norm_channels_per_group': 16, 'norm_affine': True, 'reduction': 'mean', 'loss_weight': 1.0, 'prior_prob': 0.01} 2024-11-24 21:52:04.854 | INFO | nndet.arch.heads.classifier:init_weights:215 - Init classifier weights: prior prob 0.01 2024-11-24 21:52:04.861 | INFO | nndet.ptmodule.retinaunet.base:_build_head_regressor:564 - Building:: regressor GIoURegressor: {'num_convs': 1, 'norm_channels_per_group': 16, 'norm_affine': True, 'reduction': 'sum', 'loss_weight': 1.0, 'learn_scale': True} 2024-11-24 21:52:04.873 | INFO | nndet.arch.heads.regressor:build_scales:150 - Learning level specific scalar in regressor 2024-11-24 21:52:04.873 | INFO | nndet.arch.heads.regressor:init_weights:196 - Overwriting regressor conv weight init 2024-11-24 21:52:04.883 | INFO | nndet.ptmodule.retinaunet.base:_build_head:602 - Building:: head DetectionHeadHNMNative: {} sampler HardNegativeSamplerBatched: {'batch_size_per_image': 32, 'positive_fraction': 0.33, 'pool_size': 20, 'min_neg': 1} 2024-11-24 21:52:04.883 | INFO | nndet.core.boxes.sampler:__init__:235 - Sampling hard negatives on a per batch basis 2024-11-24 21:52:04.883 | INFO | nndet.ptmodule.retinaunet.base:_build_segmenter:638 - Building:: segmenter DiCESegmenterFgBg {'dice_kwargs': {'batch_dice': True}} 2024-11-24 21:52:04.884 | INFO | nndet.losses.segmentation:__init__:108 - Running batch dice True and do bg False in dice loss. 2024-11-24 21:52:04.884 | INFO | nndet.ptmodule.retinaunet.base:from_config_plan:421 - Model Inference Summary: detections_per_img: 100 score_thresh: 0 topk_candidates: 10000 remove_small_boxes: 0.01 nms_thresh: 0.6
What's the meaning of this?，How should I use this post-processing

[partha-ghosh]

Hi @GarryJAY502,
As the message said, the post-processing step is not implemented in the current version and is kept as a placeholder for future work. So you can ignore the message for now.

Best,
Partha