OrgLion-ML
/
openface
mirror of https://github.com/cmusatyalab/openface.git
1
0
Fork 0
Code Issues Releases Wiki Activity

Docs: Attempt to fix images.

This commit is contained in:
Brandon Amos 2015-11-02 07:58:51 -05:00
parent ccfc9ecaa8
commit ae61278fb2
3 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
docs/accuracy.md
View File

@ -10,7 +10,7 @@ We obtain a mean accuracy of 0.8483 ± 0.0172 with an AUC of 0.923.
For comparison, training with Google-scale data results in an
accuracy of .9963 ± 0.009.
![](../../images/nn4.v1.lfw.roc.png)
![](../images/nn4.v1.lfw.roc.png)
This can be generated with the following commands from the root `openface`
directory, assuming you have downloaded and placed the raw and

2
docs/models.md
View File

@ -34,4 +34,4 @@ Our `nn4.v1` model is from epoch 177.
The LFW section above shows that this model obtains a mean
accuracy of 0.8483 ± 0.0172 with an AUC of 0.923.
![](../../images/nn4.v1.loss.png)
![](../images/nn4.v1.loss.png)

8
docs/visualizations.md
View File

@ -7,11 +7,11 @@ in the training and testing dataset with the most images.
**Training**
![](../../images/train-tsne.png)
![](../images/train-tsne.png)
**Testing**
![](../../images/val-tsne.png)
![](../images/val-tsne.png)
These can be generated with the following commands from the root
`openface` directory.
@ -37,5 +37,5 @@ The following shows the first 39 filters of the
first convolutional layer on two images
of John Lennon.
![](../../images/nn4.v1.conv1.lennon-1.png)
![](../../images/nn4.v1.conv1.lennon-2.png)
![](../images/nn4.v1.conv1.lennon-1.png)
![](../images/nn4.v1.conv1.lennon-2.png)