Technion - Machine Learning and Optimization (097209)
Yotam Martin, Gal Goldstein
✓ sign means we built the file from scratch
if no sign the code is from github repos and adjusted by us
.
├── advanced_model_flat.py ----------------------------- ✓ train and inference flat models (breed)
├── advanced_model_hierarchical.py --------------------- ✓ train and inference hierarchical models (breed)
├── basic_model_baseline.py ---------------------------- ✓ train and inference KNN, SVM classifiers
├── basic_model_structured.py -------------------------- ✓ inference structured FCN+CRF-RNN model
├── crfrnn_keras --------------------------------------- github repo for training CRF-RNN
│ ├── cpp
│ │ ├── Makefile
│ │ ├── high_dim_filter.cc
│ │ ├── modified_permutohedral.cc
│ │ └── modified_permutohedral.h
│ ├── crfrnn_layer.py
│ ├── crfrnn_model.py
│ ├── high_dim_filter_loader.py
│ ├── test_gradients.py
│ └── util.py
├── data_advanced_model.csv ---------------------------- ✓ paths to images for advanced part models (windows)
├── data_advanced_model_linux.csv ---------------------- ✓ paths to images for advanced part models (linux)
├── data_basic_model.csv ------------------------------- ✓ paths to images for basic part models (windows)
├── data_basic_model_linux.csv ------------------------- ✓ paths to images for basic part models (linux)
├── neural_structured_learning_adversarial_examples.py - ✓ generate examples for Figure 4 in report
├── neural_structured_learning_model.py ---------------- ✓ train and inference NSL models (creative part)
├── train_CRF-RNN -------------------------------------- github repo for training CRF-RNN
│ ├── LICENSE
│ ├── README.md
│ ├── TVG_CRFRNN_COCO_VOC_TEST_3_CLASSES.prototxt
│ ├── TVG_CRFRNN_COCO_VOC_TRAIN_3_CLASSES.prototxt
│ ├── convert_labels.py
│ ├── crfasrnn.py
│ ├── data2lmdb.py
│ ├── filter_images.py
│ ├── loss_from_log.py
│ ├── resume_training.py
│ ├── solver.prototxt
│ ├── test_model.py
│ ├── train.py
│ └── utils.py
└── utils.py ------------------------------------------- ✓ aux funcs for augmentations and .csv files