Random Ferns Research Articles

Comparison of random forest, random ferns and support vector machine for eye state classification

Eye state estimation has a wide range of potential applications, such as human-computer interaction, driver fatigue monitoring system and facial expression recognition. The commonly used feature-ba...

Robust Face Alignment Under Occlusion via Regional Predictive Power Estimation

Face alignment has been well studied in recent years, however, when a face alignment model is applied on facial images with heavy partial occlusion, the performance deteriorates significantly. In this paper, instead of training an occlusion-aware model with visibility annotation, we address this issue via a model adaptation scheme that uses the result of a local regression forest (RF) voting method. In the proposed scheme, the consistency of the votes of the local RF in each of several oversegmented regions is used to determine the reliability of predicting the location of the facial landmarks. The latter is what we call regional predictive power (RPP). Subsequently, we adapt a holistic voting method (cascaded pose regression based on random ferns) by putting weights on the votes of each fern according to the RPP of the regions used in the fern tests. The proposed method shows superior performance over existing face alignment models in the most challenging data sets (COFW and 300-W). Moreover, it can also estimate with high accuracy (72.4% overlap ratio) which image areas belong to the face or nonface objects, on the heavily occluded images of the COFW data set, without explicit occlusion modeling.

Robust Object Tracking With Reacquisition Ability Using Online Learned Detector

Long term tracking is a challenging task for many applications. In this paper, we propose a novel tracking approach that can adapt various appearance changes such as illumination, motion, and occlusions, and owns the ability of robust reacquisition after drifting. We utilize a condensation-based method with an online support vector machine as a reliable observation model to realize adaptive tracking. To redetect the target when drifting, a cascade detector based on random ferns is proposed. It can detect the target robustly in real time. After redetection, we also come up with a new refinement strategy to improve the tracker's performance by removing the support vectors corresponding to possible wrong updates by a matching template. Extensive comparison experiments on typical and challenging benchmark dataset illustrate a robust and encouraging performance of the proposed approach.

Robust facial landmark localization using classified random ferns and pose-based initialization

Facial landmark localization on images, which are captured in the wild under uncontrolled conditions, is a challenge work. Compared with localization methods using deformable models, the cascaded regression models obtain much better performance in terms of accuracy and efficiency for those unseen images. However, such regression approach is initialization dependent. In this paper, we propose a new random fern constructing method by introducing classification analysis. A robust pose-based initialization method is proposed by computing the pose similarity between the estimated image and initial shapes. The proposed method is evaluated on two “in the wild” databases, LFPW and HELEN, which are collected from websites and largely various in pose, illumination, expression and occlusion. The experimental results have demonstrated that the proposed method can significantly enhance the accuracy and robustness on facial landmark localization.

Prostate boundary segment extraction using cascaded shape regression and optimal surface detection.

In this paper, we proposed a new method (CSR+OSD) for the extraction of irregular open prostate boundaries in noisy extracorporeal ultrasound image. First, cascaded shape regression (CSR) is used to locate the position of prostate boundary in the images. In CSR, a sequence of random fern predictors are trained in a boosted regression manner, using shape-indexed features to achieve invariance against position variations of prostate boundaries. Afterwards, we adopt optimal surface detection (OSD) to refine the prostate boundary segments across 3D sections globally and efficiently. The proposed method is tested on 162 ECUS images acquired from 8 patients with benign prostate hyperplasia. The method yields a Root Mean Square Distance of 2.11±1.72 mm and a Mean Absolute Distance of 1.61±1.26 mm, which are lower than those of JFilament, an open active contour algorithm and Chan-Vese region based level set model, respectively.

Visual tracking with randomly projected ferns

In this paper, a novel visual tracking method based on random ferns and random projection has been proposed. Instead of a binary comparison in the standard random ferns method, a subtraction is used in the proposed method. Different subtraction results are projected into a single real value by the random projection. The real value from the subtraction is more informative than a binary comparison, thus less subtractions are needed. More importantly, the proposed method has a constant memory resource requirement independent from the number of subtractions in each fern, whereas in a standard random ferns method, the requirement is an exponential relationship with the number of binary comparisons. Besides low memory requirement, our approach is also computational efficient. Experimental results demonstrate that the proposed method performs better than the standard random ferns method and some other methods, especially in memory and computing resources requirements, which are of vital importance in some embedded systems.

Robust vision tracking by online random ferns and template library

In this paper, we proposed a robust tracking algorithm with an appearance model based on random ferns and template library. We adopt random Gaussian difference to generate binary features which depend on two randomly selected points and their corresponding Gaussian blur kernels. Semi-naive Bayes based random ferns are adopted as the discriminative model, and a template library including both positive templates and negative templates is used as generative model, the co-training of both discriminative and generative models gives our tracker the ability to separate foreground and background samples accurately. Besides, we also come up with a fragment based method which combines global ferns and local ferns to handle the occlusion problem. Experimental results demonstrated that the proposed algorithm performs well in terms of accuracy and robustness.

Online Pedestrian Tracking with Kalman Filter and Random Ferns

Object tracking is one of the most important components in numerous applications of computer vision. Much progress has been made in recent years. The Tracking-Learning-Detection (TLD) algorithm achieves excellent performance on a set of challenging video sequences. However, classical TLD algorithm fails to track non-rigid pedestrian as complicated appearance, varying viewpoints, shape changes and occlusions. In this paper, we follow the TLD and propose a robust KMD framework which consists of the Kalman filter tracker, random ferns pedestrian detector and the pedestrian model. During the tracking process the tracker and detector are complementary: the Kalman filter tracker predicts the motion, the pedestrian detector searches the best-match appearance and the pedestrian model combines performance of both to determine final result and generate new samples for learning. Experimental results show our framework comparatively improves performance for pedestrian tracking in surveillance videos.

Robustness of Random Forest-based gene selection methods

BackgroundGene selection is an important part of microarray data analysis because it provides information that can lead to a better mechanistic understanding of an investigated phenomenon. At the same time, gene selection is very difficult because of the noisy nature of microarray data. As a consequence, gene selection is often performed with machine learning methods. The Random Forest method is particularly well suited for this purpose. In this work, four state-of-the-art Random Forest-based feature selection methods were compared in a gene selection context. The analysis focused on the stability of selection because, although it is necessary for determining the significance of results, it is often ignored in similar studies.ResultsThe comparison of post-selection accuracy of a validation of Random Forest classifiers revealed that all investigated methods were equivalent in this context. However, the methods substantially differed with respect to the number of selected genes and the stability of selection. Of the analysed methods, the Boruta algorithm predicted the most genes as potentially important.ConclusionsThe post-selection classifier error rate, which is a frequently used measure, was found to be a potentially deceptive measure of gene selection quality. When the number of consistently selected genes was considered, the Boruta algorithm was clearly the best. Although it was also the most computationally intensive method, the Boruta algorithm’s computational demands could be reduced to levels comparable to those of other algorithms by replacing the Random Forest importance with a comparable measure from Random Ferns (a similar but simplified classifier). Despite their design assumptions, the minimal optimal selection methods, were found to select a high fraction of false positives.

Invariant Hough Random Ferns for Object Detection and Tracking

This paper introduces an invariant Hough random ferns (IHRF) incorporating rotation and scale invariance into the local feature description, random ferns classifier training, and Hough voting stages. It is especially suited for object detection under changes in object appearance and scale, partial occlusions, and pose variations. The efficacy of this approach is validated through experiments on a large set of challenging benchmark datasets, and the results demonstrate that the proposed method outperforms state-of-the-art conventional methods such as bounding-box-based and part-based methods. Additionally, we also propose an efficient clustering scheme based on the local patches’ appearance and their geometric relations that can provide pixel-accurate, top-down segmentations from IHRF back-projections. This refined segmentation can be used to improve the quality of online object tracking because it avoids the drifting problem. Thus, an online tracking framework based on IHRF, which is trained and updated in each frame to distinguish and segment the object from the background, is established. Finally, the experimental results on both object segmentation and long-term object tracking show that this method yields accurate and robust tracking performance in a variety of complex scenarios, especially in cases of severe occlusions and nonrigid deformations.

RFerns: An Implementation of the Random Ferns Method for General-Purpose Machine Learning

Random ferns is a very simple yet powerful classification method originally introduced for specific computer vision tasks. In this paper, I show that this algorithm may be considered as a constrained decision tree ensemble and use this interpretation to introduce a series of modifications which enable the use of random ferns in general machine learning problems. Moreover, I extend the method with an internal error approximation and an attribute importance measure based on corresponding features of the random forest algorithm. I also present the R package rFerns containing an efficient implementation of this modified version of random ferns.

Robust object tracking using enhanced random ferns

This paper presents a method to address the problem of long-term robust object tracking in unconstrained environments. An enhanced random fern is proposed and integrated into our tracking framework...

Lightweight generic random ferns for multi‐target augmented reality on mobile devices

Proposed use lightweight generic random ferns (LGRF), a fast keypoint classifier designed for multi-target augmented reality (AR) on mobile devices. LGRF uses binary features of image patches for both object recognition and keypoint matching of multiple objects, and stores probabilities in a single bit representation to reduce memory requirements. As a result, LGRF can perform simultaneous object recognition and keypoint matching in real time with low memory consumption, making it suitable for multi-target AR on mobile devices.

Perspective-SIFT: An efficient tool for low-altitude remote sensing image registration

This paper presents an automated image registration approach that is robust to perspective distortions. State-of-the-art method affine-SIFT uses affine transform to simulate various viewpoints to increase the robustness of registration. However, affine transformation does not follow the process by which real-world images are formed. To solve this problem, we propose a perspective scale invariant feature transform (PSIFT) that uses homographic transformation to simulate perspective distortion. As for ASIFT, PSIFT is based on the scale invariant feature transform (SIFT) and has a two-resolution scheme, namely a low-resolution phase and a high-resolution phase. The low-resolution phase of PSIFT simulates several image views following a perspective transformation by varying two camera axis orientation parameters. Given those simulated images, SIFT is then used to extract features and find matches among them. In the high-resolution phase, the perspective transformations which lead the largest number of matches in the low-resolution stage are selected to generate SIFT features on the original images. Experimental results obtained on three categories of low-altitude remote sensing images and Morel–Yu's dataset show that PSIFT outperforms significantly the state-of-the-art ASIFT, SIFT, Random Ferns, Harris-Affine, MSER and Hessian Affine, especially when images suffer severe perspective distortion.

Visual Tracking Using Combining Motion Constraint Model and Online Multiple Instance Boost Random Ferns

Video object tracking is essential algorithm for computer vision applications. An object tracking algorithm using combining motion constraints model and online multiple instance boost random ferns is proposed, which use IIR filter to obtain online learning for random ferns, and the random ferns are selected by online multiple instance boosting to construct classifier of online multiple instance boost random ferns. To reduce effects of tracking error accumulation, object motion constraint model is constructed to constrain the results classified by online multiple instance boost random ferns to locate object correctly, and construct positive and negative set to online update the classifier. The experiment shows that the proposed method achieves competitive detection results, which are comparable with state-of-the-art methods.

Bootstrapping Boosted Random Ferns for discriminative and efficient object classification

In this paper we show that the performance of binary classifiers based on Boosted Random Ferns can be significantly improved by appropriately bootstrapping the training step. This results in a classifier which is both highly discriminative and computationally efficient and is particularly suitable when only small sets of training images are available.During the learning process, a small set of labeled images is used to train the boosting binary classifier. The classifier is then evaluated over the training set and warped versions of the classified and misclassified patches are progressively added into the positive and negative sample sets for a new re-training step. In this paper we thoroughly study the conditions under which this bootstrapping scheme improves the detection rates. In particular we assess the quality of detection both as a function of the number of bootstrapping iterations and the size of the training set. We compare our algorithm against state-of-the-art approaches for several databases including faces, cars, motorbikes and horses, and show remarkable improvements in detection rates with just a few bootstrapping steps.

2A1-M15 Enhancing Localization Using Random Ferns Based Vision and Multi-Robot Collaboration(Localization and Mapping)

Humanoid robots sensory signals typically suffer from noise. In the typical case of indoor environement, small obstacles like carpets, books, or wires can make the odomtry error degenerate which eventually results in severe inaccuracies during the localization process. In this paper, we describe a landmark based multi-robot localization architecture which handles robustly the self-localization problem of a team of small humanoid robots. The landmark detection under partial occlusion and affine transformations takes advantage of the real-time capabilities of a random ferns based vision system. The observations of a single robot and those of cooperating parteners are merged through a particle filter-based method. In our approach, the abslute localization of every single robot is achieved in a robot-centric way. The relative localization is kept by a remote machine. Since all the team data is interfaced via the remote machine, every robot can act independently from the rest of the robot network.

基于随机蕨的光电成像末端制导目标初始化

基于随机蕨的光电成像末端制导目标初始化