Reference Papers and Github

• CSSR
  • Class Specific Semantic Reconstruction for Open Set Recognition [TPAMI 2022]
  • Official PyTorch implementation of Class Specific Semantic Reconstruction for Open Set Recognition.
• CAC
  • Class Anchor Clustering: a Distance-based Loss for Training Open Set Classifiers
  • Official PyTorch implementation of Class Anchor Clustering: a Distance-based Loss for Training Open Set Classifiers

---

0. Setup environment Prepare dataset

python version 3.7.11

$ pip install -r requirements.txt

Before training, please setup dataset directories in dataset.py:

DATA_PATH = ''          # path for cifar10, svhn
TINYIMAGENET_PATH = ''  # path for tinyimagenet
LARGE_OOD_PATH = ''     # path for ood datasets, e.g., iNaturalist in imagenet experiment
IMAGENET_PATH = ''      # path for imagenet-1k datasets

in case of a tinyimagenet

$ sh tinyimagenet.sh

---

1. Train

• CSSR
  • To train models from scratch, run command:

python main.py --gpu 0 --ds {DATASET} --config {MODEL} --save {SAVING_NAME} --method cssr --use_neck 

• CAC
  • To train models from scratch, run command:

python main.py --gpu 0 --ds {DATASET} --config {MODEL} --save {SAVING_NAME} --method cssr --use_neck --transfer_learning

Command options:

• DATASET: Experiment configuration file, specifying datasets and random splits, e.g., ./exps/$dataset/spl_$s.json.
• MODEL: OSR model configuration file, specifying model parameters, e.g., ./configs/$model/$dataset.json., ./configs/$model/$cac.json. $model includes linear/pcssr/rcssr, which corresponds to the baseline and the proposed model.
• store true ArgsParser
  • --head_only: finetunes only the regressor and the classifier(AutoEncoder)
  • --use_neck: Add a neck of PANet between backbone network and classifier(AutoEncoder)
  • --transfer_learning: Only use this feature for transfer learning

---

2. Evaluation

• CSSR
  • Add --test on training commands to restore and evaluate a pretrained model on specified data setup, e.g.,

python main.py --gpu 0 --ds {DATASET} --config {MODEL} --save {SAVING_NAME} --method cssr --use_neck --test

• CAC
  • Add --test on training commands to restore and evaluate a pretrained model on specified data setup, e.g.,

python main.py --gpu 0 --ds {DATASET} --config {MODEL} --save {SAVING_NAME} --method cssr --use_neck --test

---

3. Result feature map

• ground truth
  • Coloring based on ground truth.
• predictions
  • Coloring based on model's predictions before openset recognition
• openset recognition
  • Coloring based on openset recognition