Hello!
In my researches I work with face detection compare methods (on Viola-Jones example). But I still have not found face detection metrics survey (or something like), or object detection methods survey. I use F-measure (Cornelis J. van Rijsbergen).
May you give some advice about metrics? Or some links to surveys.
Article Selection of Viola–Jones algorithm parameters for specific conditions
Article Algorithm for optimization of Viola–Jones object detection f...
Conference Paper Some cases of optimization face detection methodes on image ...
Hey there,
Basically, you need to generate a ROC/DET curve from the detections and compute the AUC. This is not a simple process to explain here, however, you can find details regarding this process in the papers:
"Pedestrian Detection: An Evaluation of the State of the Art"
"A Study of Filtering Approaches for Sliding Window Pedestrian Detection"
There is a toolkit for pedestrian detection evaluation (using sliding window), where you can change it for face detection.
Link to the toolkit:
"http://www.vision.caltech.edu/Image_Datasets/CaltechPedestrians/"
This image shows the results of some face detectors I have implemented. The values next to method's name indicate the AUC, where low values are better. This figure has been generated from the toolkit I mentioned above.
Dear Alexey Egorov ,
Look the link, maybe useful.
https://pdfs.semanticscholar.org/d051/ba0a904e4b4c45a2af145aa29b8490bbbc5c.pdf
http://isg.cs.tcd.ie/cosulliv/Pubs/LeveyWSCG00.pdf
Regards, Shafagat
For face face detection evaluation, the ROC curve analysis is the most used approach. About ROC curves see this paper: Article Introduction to ROC analysis
ROC curve is the best one. In MATLAB help there examples are given
ROC curves are the most used approach. See this link: http://vis-www.cs.umass.edu/fddb/
- Badges
- Science topic
- Similar topics
- Computer Science
- Images