Uploaded bySmriti Tikoo
face detection
This document presents a literature review and proposed work plan for face recognition using a back propagation neural network. It summarizes the Viola-Jones face detection algorithm which uses Haar features and an integral image for real-time detection. The algorithm has high detection rates with low false positives. Future work will apply back propagation neural networks to extract features and recognize faces from a database of facial images in order to build a facial recognition system.
More Related Content
face detection
- 1. Face Recognition UsingBack Propagation Neural Network Supervised by Made By Dr. Nitin Malik Smriti Tikoo ECE Dept. 14-ECP-015 Mtech-ECE
- 2. Agenda • Literature Survey •Face detection • Face detection algorithms • Viola Jones algorithm • Flowchart of algorithm • Feature types and Evaluation • Haar Features • Creating an integral image • Advantages • Limitations • Faces and features detected • Conclusion • Work Plan for next semester • References
- 3. Literature Survey • BackPropagation-most popular and simplified training algorithm of feed forward neural network. • A supervised learning algorithm. • Uses sigmoid activation function. • Prescribes a gradient based weight update rule to learn a pair of input / output vectors in a feed forward neural network. • Consists of propagation feed forward and back propagation of activation and error function. • In feed forward the weights and biases are fixed is fixed while in backward they change in accordance with error correction rule. • By using Local minimum detector algorithm we can avoid trapping in local minima problem.
- 4. Face Detection • Facedetection /recognition is employed for surveillance so as to identify or verify a face from the available facial data base. • Some facial algorithms identify by doing facial feature extraction , or by analyzing relative position , size and or shape of eyes , cheekbones etc. • These features are then used to search images with matching features.
- 5. • Other algorithmsnormalize a gallery of face images and then compress the face data , saving only the data useful in image recognition. • A probe is then compared with face data .
- 6. Face Recognition Algorithms •Viola Jones Face detection algorithm • Local Binary Pattern • Adaptive boost algorithm • Principal Component Analysis • Independent Component Analysis • Eigen Space based algorithm • Kernel methods • Hidden Markov Model
- 7. Viola Jones Algorithm •The Viola–Jones object detection framework is the first object detection framework to provide competitive object detection rates in real-time proposed in 2001 by Paul Violaand Michael Jones. • Although it can be trained to detect a variety of object classes, it was motivated primarily by the problem of face detection. • This algorithm is implemented in OpenCV as cvHaarDetectObjects().k
- 8. Feature Types andEvaluation • The characteristics of Viola–Jones algorithm which make it a good detection algorithm are: • Robust – very high detection rate (true-positive rate) & very low false-positive rate always. • Real time – For practical applications at least 2 frames per second must be processed. • Face detection only (not recognition) - The goal is to distinguish faces from non-faces (detection is the first step in the recognition process). • The algorithm has four stages: • Haar Feature Selection • Creating an Integral Image • Ada boost Training • Cascading Classifiers
- 9. Flowchart of algorithm INPUTIMAGE HAAR FETURE SELECTION INTEGRAL IMAGE ADABOOST TRAINING CASCADING CLASSIFIERS
- 10. Haar Features • HaarFeatures – All human faces share some similar properties. These regularities may be matched using Haar Features. • A few properties common to human faces: • The eye region is darker than the upper-cheeks. • The nose bridge region is brighter than the eyes. • Composition of properties forming matchable facial features: • Location and size: eyes, mouth, bridge of nose • Value: oriented gradients of pixel intensities
- 11. Continue…. • The fourfeatures matched by this algorithm are then sought in the image of a face (shown at left). • Rectangle features: • Value = Σ (pixels in black area) - Σ (pixels in white area) • Three types: two-, three-, four-rectangles, Viola & Jones used two-rectangle features • For example: the difference in brightness between the white &black rectangles over a specific area • Each feature is related to a special location in the sub- window
- 12.
- 13. Creating an integralimage • An image representation called the integral image evaluates rectangular features in constant time, which gives them a considerable speed advantage over more sophisticated alternative features. • Because each feature's rectangular area is always adjacent to at least one other rectangle, it follows that any two-rectangle feature can be computed in six array references, any three-rectangle feature in eight, and any four-rectangle feature in nine. • The integral image at location (x,y), is the sum of the pixels above and to the left of (x,y), inclusive.
- 14. Advantages • Extremely fastfeature computation • Efficient feature selection • Scale and location invariant detector • Instead of scaling the image itself (e.g. pyramid-filters), we scale the features. • Such a generic detection scheme can be trained for detection of other types of objects (e.g. cars, hands)
- 15. Limitations • Detector ismost effective only on frontal images of faces • It can hardly cope with 45° face rotation both around the vertical and horizontal axis. • Sensitive to lighting conditions • We might get multiple detections of the same face, due to overlapping sub-windows.
- 16.
- 17.
- 19. Conclusion • Facial detectionis influenced by clarity of the image , colored or black and white image . • It can only support frontal detection of images. • The training does takes a lot of time in order to separate a negative face from a positive face. • Using Adaptive boost algorithm and cascading helps in faster detection.
- 20. Work Plan forNext semester • Face Recognition • Feature extraction and recognition • Facial image of each person to be collected in n number of samples for testing and comparison. • Using Back propagation Neural Network to contemplate the facial recognition system • Using Neural Network Fitting Tool for network generalization and checking the performance in terms of mean square errors. • Training to be performed again and again to reduce the error.
- 21. References • N. Revathy,“Face Recognition using Back Propagation ANN’s”, International Journal of Advance Engineering Technology”, Vol-3 issue 1, 2012. • Tej Pal Singh TIT Bhopal ,“Face recognition using feed forward back propagation neural network“, International Journal of Innovative Research in Technology & Science(IJIRTS) , vol-1 Number1,ISSN:2321-1156 , March 30,2013 • http://utarcvis.blogspot.in/2007/09/robust-real-time- face-detection-by-paul.html. • https://www.youtube.com/watch?v=WfdYYNamHZ8. • https://en.wikipedia.org/wiki/Viola%E2%80%93Jones_ object_detection_framework.