Facial occlusion, such as sunglasses, scarf, mask, etc., is one critical factor that affects the performance of face recognition. Unfortunately, faces with occlusion are quite common in the real world, especially in the uncooperative scenario. In recent years, regression analysis becomes a hotspot for dealing with face recognition under different illuminations and facial occlusions. The basic idea of regression analysis is to recover clean images from degraded images or occluded images by using clean training samples.
Then the reconstructed images are used for face recognition. However, the noise would be introduced in the recovery procedure. So whether reconstructed images help face recognition is still worth studying. Note that the residual image which is a difference between the raw and reconstructed image containing most of the occluded information.
We can use it for occlusion detection. In this paper we make two contributions: i) we present a new occlusion detection method by combining the information of both raw image and residual image; ii) we empirically show that using the non-occluded part for face recognition has a better result than using a reconstructed image.
Data gathering Data collection is the process of collecting and evaluating information in a defined systematic fashion on variables of interest, which helps one to answer specified research questions, test hypotheses, and analyze results.
Training data is an initial collection of data used to help a program understand how to learn and generate sophisticated results by implementing technologies such as neural networks. It can be supplemented by subsequent data sets called validation sets and test sets.
Face Recognition Recognition of the face is a way of recognizing or confirming an individual's identity using their face. It is possible to use face recognition systems to recognize people in pictures, videos, or in real-time. In order to distinguish persons during police stops, law enforcement can also use mobile devices.
Face Classification The final step uses extracted FacialFeature s to perform face recognition or classification (determining who the face is) or classification (determining certain face characteristics).
Face detection In a number of applications, face recognition is a computer technology that detects human faces in digital images. In a visual scene, face recognition also refers to the psychological mechanism by which humans identify and attend to faces.
Haar cascade algorithm
supported operating system
Cost and technology
|Project Details contact - +91 9964716807|
|project title||face recognition with occlusion using haar cascade algorithm|
|project type||Final year project.|
|Project price||₹ 12000 INR|
|Project discount||₹ 2000 INR|
|Final price||₹ 10000 INR|
|Documentation||charges extra cost for documentation for all project|
By utilizing the achievement of the NMR method, we proposed a robustness method for the detection of sunglasses and scarves. Yet, there are many other occlusions such as hats, beards, etc., our method can still be extended to a more general condition by comparing the difference between the raw images and the reconstructed images. On the other hand, reconstructed images that are created by using the existing regression analysis methods are not as accurate as we thought. Aliasing and noise may be introduced during the procedure of reconstruction. Although the structure of the face is more complete, some useful information has been destroyed and the incidental information has been increased. The non-occluded facial region of the occluded face can still provide more accurate information than recovering face images. The state-of-the-art regression-based methods for face recognition with occlusion still need to be improved. How to improve non-occluded face recognition with an advanced classifier will be another research direction.