Loading...
Please wait, while we are loading the content...
Similar Documents
Faces retouched
| Content Provider | Indraprastha Institute of Information Technology, Delhi |
|---|---|
| Author | Bharati, Aparna |
| Abstract | People generally retouch their photos and these altered photos are used for uploading on social media, photo sharing websites, and even for getting identication cards where the standards are not strictly followed. These alterations or retouching is mostly done using softwares and is thus, very different from disguise and makeup and is relatively less explored. This research studies the effect of these digital alterations on the performance of automatic face recognition. We first introduce two face image databases with unaltered and retouched images. To show the implications of face retouching, recognition experiments are performed on these databases and the results show that retouching reduces face recognition accuracy. This is an interesting observation and our assertion is that the accuracy difference occurs because applying retouching on gallery and probe images minimizes the inter-class differences. We also discuss the results of a popular image tampering/alteration detection algorithm on these two databases as well as on the LFW face database. We propose a novel algorithm to detect retouched faces. The algorithm uses discriminative DBM on prominent face patches to predict whether the image is unaltered or retouched. |
| File Format | |
| Language | English |
| Access Restriction | Authorized |
| Subject Keyword | Retouching Face detection Classification accuracy Recognition FACEVACS OpenBR Deep belief networks Histogram of oriented gradients |
| Content Type | Text |
| Educational Degree | Bachelor of Technology (B.Tech.) |
| Subject | Data processing & computer science |
