Details: Facial Kinship Verification (FKV) is a vision based research challenge in computer vision field. It is vision based research challenge. This research has potential applications, finding missing children using facial images, criminal investigations, border control, customs, social media analysis and album management. In general, children resemble their parents, this is the main hypothesis behind FKV research. This is a binary classification problem.

In the demo video, you can see two test samples are passing to find whether these two are blood related or not. This is a main objective of the FKV.

On this research topic CV Lab has published 10-15 peer reviewed articles. Further, Children based kinship verification and kin pair image generation research is under progress.

Details: Action recognition is a fundamental task in the computer vision community that recognizes human actions based on the complete action execution in a video and action prediction is a before-the-fact video understanding task and is focusing on the future state. In some real-world scenarios (e.g. vehicle accidents and criminal activities) intelligent machines do not have the luxury of waiting for the entire action execution before having to react to the action contained in it.

From the demo video you can see the results of different action recognition along with direction of our proposed model.

Details: Image forensics deals with any kind of image manipulation which aims to deliver deceptive information by changing the image graphic content. The detection of image forgery without any prior information along with the detection of computer generated fake images come under the image forensics techniques. Image forgery detection categorizes the image forgery mostly in three parts: copy-move forgery, inpainting or removal and splicing. In copy-move forgery, an object or region is duplicated in the same image. If new object is inserted from different image source, this kind of forgery is called as splicing. In image inpainting any object is removed, and the empty region is filled by extending the background.