Skip to content

Latest commit

 

History

History
43 lines (35 loc) · 2.63 KB

code_structure.md

File metadata and controls

43 lines (35 loc) · 2.63 KB
Raw

Code structure

configs/, Includes the config files to run models

data/, Storing datasets (eg. data/scannet/ or data/ARKitScenes)

config_loader.py, Defines all hyper-parameters of the model

dataprocessing

  • dataprocessing/augmentation.py, Defines augmentation code
  • dataprocessing/scannet.py, Reads on train/test/val scenes of scannet
  • dataprocessing/arkitscenes.py, Reads on train/val scenes of Arkitscenes
  • dataprocessing/s3dis.py, Reads on train/val scenes of S3DIS

models

  • models/dataloader.py, Reads and preprocesses data and prepare tensor batches

    • class ScanNet, Reads and preprocesses Scannet scenes
      • approx_association(), Finds the associations of points using GT bounding boxes
      • getitem(), Preprocesses the scenes, returns model inputs and labels
    • class ARKitScenes, Reads and preprocess ArkitScenes scenes
      • approx_association(), Finds the associations of points using GT bounding boxes
      • getitem(), Preprocesses the scenes, returns model inputs and labels
    • class S3DIS, Reads and preprocess S3dis scenes
      • approx_association(), Finds the associations of points using GT bounding boxes
      • getitem(), Preprocesses the scenes, returns model inputs and labels
    • collate_fn, Collates preprocessed scenes into tensor batches

  • models/detection_net.py, Defines the network

    • class SelectionNet, Define the main network and network heads
      • detection2mask(), Converts box proposals into final instance mask
      • get_prediction(), Gets prediction from the network heads

  • models/evaluation.py, Evaluates Scannet and ArkitScenes predictions. Can be run with: python models/evaluation.py --config configs/[config_name].txt

    • arkitscenes_eval(), Approximates oriented bounding boxes from instance predictions and computes detection quality using the AP score
    • scannet_eval(), Computes Scannet prediction scores in terms of AP, AP50 and AP25

  • models/iou_nms.py, Defines the Non-Maximum Clustering clustering

    • NMS_clustering(), Non-Maximum Clustering algorithm (as in Sec.3 and Sec. 4 in the main paper)

  • models/resnet.py, Some utilities for making the U-Net model

  • models/training.py, Defines the training code, can be run witch: python models/training.py --config configs/[config_name].txt

  • models/model.py, Defines and computes the losses for each epoch

    • compute_loss_detection(), Compute each loss and the weighted joint losses for the network optimization

utils/, Contains some low-level utilities