Skip to content

Commit

Permalink
update READMEs
Browse files Browse the repository at this point in the history
  • Loading branch information
@mvp18
mvp18 committed May 2, 2020
1 parent d5e0b51 commit 494f953
Show file tree
Hide file tree
Showing 3 changed files with 37 additions and 37 deletions.
20 changes: 10 additions & 10 deletions ESZSL/README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -11,19 +11,19 @@ For testing, set mode to test and set alpha, gamma to best combination from tabl

#### Classical ZSL

| Dataset | ZSLGBU Results| Respository Results |Hyperparams from Val |
|---------|:-------------:|:-------------------:|:-------------------:|
| CUB | 53.9 | 53.94 |Alpha=3, Gamma=-1 |
| AWA1 | **58.2** | 56.80 |Alpha=3, Gamma=0 |
| AWA2 | **58.6** | 54.82 |Alpha=3, Gamma=0 |
| aPY | 38.3 | **38.56** |Alpha=3, Gamma=-1 |
| SUN | 54.5 | **55.69** |Alpha=3, Gamma=2 |
| Dataset | ZSLGBU Results| Repository Results | Hyperparams from Val |
|---------|:-------------:|:------------------:|:--------------------:|
| CUB | 53.9 | 53.94 |Alpha=3, Gamma=-1 |
| AWA1 | **58.2** | 56.80 |Alpha=3, Gamma=0 |
| AWA2 | **58.6** | 54.82 |Alpha=3, Gamma=0 |
| aPY | 38.3 | **38.56** |Alpha=3, Gamma=-1 |
| SUN | 54.5 | **55.69** |Alpha=3, Gamma=2 |

#### Generalized ZSL

|Dataset |ZSLGBU Results |||Respository Results |||Hyperparams from Val |
|--------|:-----:|:-----:|:-----:|:-----:|:----:|:-----:|:-------------------:|
| | U | S | H | U | S | H | |
|Dataset || ZSLGBU Results ||| Repository Results || Hyperparams from Val |
|--------|:-----:|:-----:|:-----:|:-----:|:----:|:-----:|:--------------------:|
| | U | S | H | U | S | H | |
| CUB | 12.6 | **63.8** | 21.0 | **14.70** | 56.53 | **23.34** |Alpha=3, Gamma=0 |
| AWA1 | **6.6** | 75.6 | **12.1** | 5.29 | **86.84** | 9.98 |Alpha=3, Gamma=0 |
| AWA2 | **5.9** | 77.8 | **11.0** | 4.04 | **88.84** | 7.72 |Alpha=3, Gamma=0 |
Expand Down
44 changes: 22 additions & 22 deletions README.md
View file
Original file line number Diff line number Diff line change
@@ -1,34 +1,34 @@
# Popular-ZSL-Algorithms
## Introduction

This repository contains implementations of 5 classical zero-shot algorithms in the usual as well as the generalized zero-shot settings using the
This repository contains implementations of 5 classical zero-shot algorithms in the usual as well as the Generalized zero-shot settings using the
`Proposed Split` and evaluation protocols outlined in
[Zero-Shot Learning - A Comprehensive Evaluation of the Good, the Bad and the Ugly (ZSLGBU)](https://arxiv.org/abs/1707.00600) by Yongqin Xian, Christoph H. Lampert, Bernt Schiele, Zeynep Akata (TPAMI 2018).
[Zero-Shot Learning - A Comprehensive Evaluation of the Good, the Bad and the Ugly](https://arxiv.org/abs/1707.00600) (**ZSLGBU**) by Yongqin Xian, Christoph H. Lampert, Bernt Schiele, Zeynep Akata (TPAMI 2018).

The original papers corressponding to the 5 algorithms are:
To the best of my knowledge, this is the **first public implementation** of SAE, ALE, SJE and DeViSE under the ZSLGBU protocol. An existing implementation of `ESZSL` can be found [here](https://github.com/sbharadwajj/embarrassingly-simple-zero-shot-learning). To this, I have also added the Generalized ZSL functionality.

- [1] SAE - [Semantic Autoencoder for Zero-Shot Learning](https://arxiv.org/abs/1704.08345).
Elyor Kodirov, Tao Xiang, Shaogang Gong.
CVPR, 2017.
## Reference Papers

- [2] ALE - [Label-Embedding for Image Classification](https://arxiv.org/abs/1503.08677).
Zeynep Akata, Florent Perronnin, Zaid Harchaoui, Cordelia Schmid.
PAMI, 2016.
The original papers corresponding to the 5 algorithms are:

- [3] SJE - [Evaluation of Output Embeddings for Fine-Grained Image Classification](https://arxiv.org/abs/1409.8403).
Zeynep Akata, Scott Reed, Daniel Walter, Honglak Lee, Bernt Schiele.
CVPR, 2015.
[1] SAE - [Semantic Autoencoder for Zero-Shot Learning](https://arxiv.org/abs/1704.08345).
Elyor Kodirov, Tao Xiang, Shaogang Gong.
CVPR, 2017.

- [4] ESZSL - [An embarrassingly simple approach to zero-shot learning](http://proceedings.mlr.press/v37/romera-paredes15.pdf).
Bernardino Romera-Paredes, Philip H. S. Torr.
ICML, 2015.
[2] ALE - [Label-Embedding for Image Classification](https://arxiv.org/abs/1503.08677).
Zeynep Akata, Florent Perronnin, Zaid Harchaoui, Cordelia Schmid.
TPAMI, 2016.

- [5] DeViSE - [DeViSE: A Deep Visual-Semantic Embedding Model](http://papers.nips.cc/paper/5204-devise-a-deep-visual-semantic-embedding-model.pdf).
Andrea Frome*, Greg S. Corrado*, Jonathon Shlens*, Samy Bengio, Jeffrey Dean, Marc’Aurelio Ranzato, Tomas Mikolov.
NIPS, 2013.
[3] SJE - [Evaluation of Output Embeddings for Fine-Grained Image Classification](https://arxiv.org/abs/1409.8403).
Zeynep Akata, Scott Reed, Daniel Walter, Honglak Lee, Bernt Schiele.
CVPR, 2015.

To the best of my knowledge, this is the first public implementation of SAE, ALE, SJE and DeViSE under the ZSLGBU protocol. An existing implementation of `ESZSL` can be found [here](https://github.com/sbharadwajj/embarrassingly-simple-zero-shot-learning). To this, I have also added the generalized ZSL functionality.
[4] ESZSL - [An embarrassingly simple approach to zero-shot learning](http://proceedings.mlr.press/v37/romera-paredes15.pdf).
Bernardino Romera-Paredes, Philip H. S. Torr.
ICML, 2015.

Each code folder has its own `README` with running instructions, results and their comparisons with those found in the [ZSLGBU](https://arxiv.org/abs/1707.00600) paper.
[5] DeViSE - [DeViSE: A Deep Visual-Semantic Embedding Model](http://papers.nips.cc/paper/5204-devise-a-deep-visual-semantic-embedding-model.pdf).
Andrea Frome*, Greg S. Corrado*, Jonathon Shlens*, Samy Bengio, Jeffrey Dean, Marc’Aurelio Ranzato, Tomas Mikolov.
NIPS, 2013.

## Setup

Expand All @@ -38,4 +38,4 @@ cd Popular-ZSL-Algorithms
bash setup.sh
```

This downloads data corressponding to the `Proposed Split` for AWA1, AWA2, CUB, SUN and aPY.
This downloads data (splits, Res101 features and class embeddings) corresponding to the `Proposed Split` for AWA1, AWA2, CUB, SUN and aPY. To know more about the individual files, refer to the `README.txt` file available inside `xlsa17` folder.
10 changes: 5 additions & 5 deletions SAE/README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -11,9 +11,9 @@ For testing, set mode to test and set alpha, gamma to best combination from tabl

#### Classical ZSL

| Dataset | ZSLGBU Results | Respository Results ||||
|---------|:--------------:|:----------------------------------------:|
| | | F->S (W) | Lambda | S->F (W^T) | Lambda |
| Dataset | ZSLGBU Results ||| Repository Results ||
|---------|:--------------:|:--------:|:------:|:----------:|:-------:|
| | | F->S (W) | Lambda | S->F (W.T) | Lambda |
| CUB | 33.3 | 39.48 | 100 | **46.70** | 0.2 |
| AWA1 | 53.0 | 51.34 | 3.0 | **59.89** | 0.8 |
| AWA2 | 54.1 | 51.66 | 0.6 | **60.51** | 0.2 |
Expand All @@ -22,9 +22,9 @@ For testing, set mode to test and set alpha, gamma to best combination from tabl

#### Generalized ZSL

|Dataset |ZSLGBU Results |||Respository Results ||||||||
|Dataset |ZSLGBU Results |||Repository Results ||||||||
|--------|:-----:|:-----:|:-----:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|
| | | | | F->S (W) ||| Lambda | S->F (W^T) ||| Lambda |
| | | | | F->S (W) ||| Lambda | S->F (W.T) ||| Lambda |
| | U | S | H | U | S | H | | U | S | H | |
| CUB | 12.6 | **63.8** | 21.0 | **14.70** | 56.53 | **23.34** | 1 | **14.70** | 56.53 | **23.34** | 1 |

Expand Down

0 comments on commit 494f953

Please sign in to comment.