This code replicates the results from the paper “Deep metric learning using Triplet network” (http://arxiv.org/abs/1412.6622).
It can train a TripletNet on any of the {Cifar10/100, STL10, SVHN, MNIST} datasets.
You can get the needed data using the following repos:
- CIFAR10/100: https://github.com/soumith/cifar.torch.git
- STL10: https://github.com/eladhoffer/stl10.torch
- SVHN: https://github.com/torch/tutorials/blob/master/A_datasets/svhn.lua
- MNIST: https://github.com/andresy/mnist
- Torch (http://torch.ch)
- "eladtools" (https://github.com/eladhoffer/eladtools) for optimizer.
- "nngraph" (https://github.com/torch/nngraph) for TripletNet configuration.
- "cudnn.torch" (https://github.com/soumith/cudnn.torch) for faster training. Can be avoided by changing "cudnn" to "nn" in models.
Available models are at the “Models” directory. The basic Model.lua was used in the paper, while NiN based models achieve slightly better results.
You can start training using:
th Main.lua -dataset Cifar10 -LR 0.1 -save new_exp_dir| Flag | Default Value | Description |
|---|---|---|
| modelsFolder | ./Models/ | Models Folder |
| network | Model.lua | Model file - must return valid network. |
| LR | 0.1 | learning rate |
| LRDecay | 0 | learning rate decay (in # samples |
| weightDecay | 1e-4 | L2 penalty on the weights |
| momentum | 0.9 | momentum |
| batchSize | 128 | batch size |
| optimization | sgd | optimization method |
| epoch | -1 | number of epochs to train (-1 for unbounded) |
| threads | 8 | number of threads |
| type | cuda | float or cuda |
| devid | 1 | device ID (if using CUDA) |
| load | none | load existing net weights |
| save | time-identifier | save directory |
| dataset | Cifar10 | Dataset - Cifar10, Cifar100, STL10, SVHN, MNIST |
| normalize | 1 | 1 - normalize using only 1 mean and std values |
| whiten | false | whiten data |
| augment | false | Augment training data |
| preProcDir | ./PreProcData/ | Data for pre-processing (means,Pinv,P) |