Skip to content

Commit

Permalink
better text
Browse files Browse the repository at this point in the history
  • Loading branch information
@liuliu
liuliu committed Dec 14, 2014
1 parent 503e01f commit 3213371
Showing 1 changed file with 3 additions and 3 deletions.
6 changes: 3 additions & 3 deletions doc/scd.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -47,7 +47,7 @@ Please make sure the evaluation mode is ellipse format.

The following graph will likely to be generated:

![DiscROC-SCD](resources/disc-roc-scd.png).
![Discrete ROC for SCD](resources/disc-roc-scd.png)

The threshold is tuned to generate about 200~300 false positives.

Expand All @@ -57,7 +57,7 @@ The BBF implementation can be evaluated too:

The generated graph:

![DiscROC-BBF](resources/disc-roc-bbf.png)
![Discrete ROC for BBF](resources/disc-roc-bbf.png)

You can compare the performance with the state of the art face detectors on
FDDB website, at about 250 false positives, this implementation at 72.93% is 4.23%
Expand Down Expand Up @@ -92,7 +92,7 @@ detection rate, where as SCD with 24x24 template and color image has 72.11%
detection rate. The BBF implementation has 65.11% detection rate at the same
false positives.

![DiscROC-SCD-Fast](resources/disc-roc-scd-fast.png)
![Discrete ROC for SCD in Fast Mode](resources/disc-roc-scd-fast.png)

How to train my own detector?
-----------------------------

0 comments on commit 3213371

Please sign in to comment.