Towards automatic object segmentation with sequential multiple views

Object segmentation is one of the fundamental steps for a number of robotic applications such as manipulation, object detection, and obstacle avoidance. This paper proposes a visual method for incorporating colour and depth information from sequential multiview stereo images to segment objects of interest from complex and cluttered environments. Rather than segmenting objects using information from a single frame in the sequence, we incorporate information from neighbouring views to increase the reliability of the information and improve the overall segmentation result. Specifically, dense depth information of a scene is computed using multiple view stereo. Depths from neighbouring views are reprojected into the reference frame to be segmented compensating for imperfect depth computations for individual frames. The multiple depth layers are then combined with color information from the reference frame to create a Markov random field to model the segmentation problem. Finally, graphcut optimisation is employed to infer pixels belonging to the object to be segmented. The segmentation accuracy is evaluated over images from an outdoor video sequence demonstrating the viability for automatic object segmentation for mobile robots using monocular cameras as a primary sensor.

Automatic object segmentation of unstructured scenes using colour and depth maps

This study presents a segmentation pipeline that fuses colour and depth information to automatically separate objects of interest in video sequences captured from a quadcopter. Many approaches assume that cameras are static with known position, a condition which cannot be preserved in most outdoor robotic applications. In this study, the authors compute depth information and camera positions from a monocular video sequence using structure from motion and use this information as an additional cue to colour for accurate segmentation. The authors model the problem similarly to standard segmentation routines as a Markov random field and perform the segmentation using graph cuts optimisation. Manual intervention is minimised and is only required to determine pixel seeds in the first frame which are then automatically reprojected into the remaining frames of the sequence. The authors also describe an automated method to adjust the relative weights for colour and depth according to their discriminative properties in each frame. Experimental results are presented for two video sequences captured using a quadcopter. The quality of the segmentation is compared to a ground truth and other state-of-the-art methods with consistently accurate results.

GrabCutSFM : how 3D information improves unsupervised object segmentation

In this paper, we present an unsupervised graph cut based object segmentation method using 3D information provided by Structure from Motion (SFM), called Grab- CutSFM. Rather than focusing on the segmentation problem using a trained model or human intervention, our approach aims to achieve meaningful segmentation autonomously with direct application to vision based robotics. Generally, object (foreground) and background have certain discriminative geometric information in 3D space. By exploring the 3D information from multiple views, our proposed method can segment potential objects correctly and automatically compared to conventional unsupervised segmentation using only 2D visual cues. Experiments with real video data collected from indoor and outdoor environments verify the proposed approach.

SeamSeg: Video Object Segmentation using Patch Seams

In this paper, we propose a technique for video object segmentation using patch seams across frames. Typically, seams, which are connected paths of low energy, are utilised for retargeting, where the primary aim is to reduce the image size while preserving the salient image contents. Here, we adapt the formulation of seams for temporal label propagation. The energy function associated with the proposed video seams provides temporal linking of patches across frames, to accurately segment the object. The proposed energy function takes into account the similarity of patches along the seam, temporal consistency of motion and spatial coherency of seams. Label propagation is achieved with high fidelity in the critical boundary regions, utilising the proposed patch seams. To achieve this without additional overheads, we curtail the error propagation by formulating boundary regions as rough-sets. The proposed approach out-perform state-of-the-art supervised and unsupervised algorithms, on benchmark datasets.

Automatic object segmentation from calibrated images

This paper addresses the problem of automatically obtaining the object/background segmentation of a rigid 3D object observed in a set of images that have been calibrated for camera pose and intrinsics. Such segmentations can be used to obtain a shape representation of a potentially texture-less object by computing a visual hull. We propose an automatic approach where the object to be segmented is identified by the pose of the cameras instead of user input such as 2D bounding rectangles or brush-strokes. The key behind our method is a pairwise MRF framework that combines (a) foreground/background appearance models, (b) epipolar constraints and (c) weak stereo correspondence into a single segmentation cost function that can be efficiently solved by Graph-cuts. The segmentation thus obtained is further improved using silhouette coherency and then used to update the foreground/background appearance models which are fed into the next Graph-cut computation. These two steps are iterated until segmentation convergences. Our method can automatically provide a 3D surface representation even in texture-less scenes where MVS methods might fail. Furthermore, it confers improved performance in images where the object is not readily separable from the background in colour space, an area that previous segmentation approaches have found challenging. © 2011 IEEE.

Learning object segmentation from video data

This memo describes the initial results of a project to create a self-supervised algorithm for learning object segmentation from video data. Developmental psychology and computational experience have demonstrated that the motion segmentation of objects is a simpler, more primitive process than the detection of object boundaries by static image cues. Therefore, motion information provides a plausible supervision signal for learning the static boundary detection task and for evaluating performance on a test set. A video camera and previously developed background subtraction algorithms can automatically produce a large database of motion-segmented images for minimal cost. The purpose of this work is to use the information in such a database to learn how to detect the object boundaries in novel images using static information, such as color, texture, and shape. This work was funded in part by the Office of Naval Research contract #N00014-00-1-0298, in part by the Singapore-MIT Alliance agreement of 11/6/98, and in part by a National Science Foundation Graduate Student Fellowship.

Foreground Object Segmentation from Binocular Stereo Video

Moving cameras are needed for a wide range of applications in robotics, vehicle systems, surveillance, etc. However, many foreground object segmentation methods reported in the literature are unsuitable for such settings; these methods assume that the camera is fixed and the background changes slowly, and are inadequate for segmenting objects in video if there is significant motion of the camera or background. To address this shortcoming, a new method for segmenting foreground objects is proposed that utilizes binocular video. The method is demonstrated in the application of tracking and segmenting people in video who are approximately facing the binocular camera rig. Given a stereo image pair, the system first tries to find faces. Starting at each face, the region containing the person is grown by merging regions from an over-segmented color image. The disparity map is used to guide this merging process. The system has been implemented on a consumer-grade PC, and tested on video sequences of people indoors obtained from a moving camera rig. As can be expected, the proposed method works well in situations where other foreground-background segmentation methods typically fail. We believe that this superior performance is partly due to the use of object detection to guide region merging in disparity/color foreground segmentation, and partly due to the use of disparity information available with a binocular rig, in contrast with most previous methods that assumed monocular sequences.

Automatic object segmentation from calibrated images

This paper addresses the problem of automatically obtaining the object/background segmentation of a rigid 3D object observed in a set of images that have been calibrated for camera pose and intrinsics. Such segmentations can be used to obtain a shape representation of a potentially texture-less object by computing a visual hull. We propose an automatic approach where the object to be segmented is identified by the pose of the cameras instead of user input such as 2D bounding rectangles or brush-strokes. The key behind our method is a pairwise MRF framework that combines (a) foreground/background appearance models, (b) epipolar constraints and (c) weak stereo correspondence into a single segmentation cost function that can be efficiently solved by Graph-cuts. The segmentation thus obtained is further improved using silhouette coherency and then used to update the foreground/background appearance models which are fed into the next Graph-cut computation. These two steps are iterated until segmentation convergences. Our method can automatically provide a 3D surface representation even in texture-less scenes where MVS methods might fail. Furthermore, it confers improved performance in images where the object is not readily separable from the background in colour space, an area that previous segmentation approaches have found challenging. © 2011 IEEE.

Deep convolutional neural networks for semantic video object segmentation

In this thesis, we propose to infer pixel-level labelling in video by utilising only object category information, exploiting the intrinsic structure of video data. Our motivation is the observation that image-level labels are much more easily to be acquired than pixel-level labels, and it is natural to find a link between the image level recognition and pixel level classification in video data, which would transfer learned recognition models from one domain to the other one. To this end, this thesis proposes two domain adaptation approaches to adapt the deep convolutional neural network (CNN) image recognition model trained from labelled image data to the target domain exploiting both semantic evidence learned from CNN, and the intrinsic structures of unlabelled video data. Our proposed approaches explicitly model and compensate for the domain adaptation from the source domain to the target domain which in turn underpins a robust semantic object segmentation method for natural videos. We demonstrate the superior performance of our methods by presenting extensive evaluations on challenging datasets comparing with the state-of-the-art methods.

Multiplicative Kernels: Object Detection, Segmentation and Pose Estimation

Object detection is challenging when the object class exhibits large within-class variations. In this work, we show that foreground-background classification (detection) and within-class classification of the foreground class (pose estimation) can be jointly learned in a multiplicative form of two kernel functions. One kernel measures similarity for foreground-background classification. The other kernel accounts for latent factors that control within-class variation and implicitly enables feature sharing among foreground training samples. Detector training can be accomplished via standard SVM learning. The resulting detectors are tuned to specific variations in the foreground class. They also serve to evaluate hypotheses of the foreground state. When the foreground parameters are provided in training, the detectors can also produce parameter estimate. When the foreground object masks are provided in training, the detectors can also produce object segmentation. The advantages of our method over past methods are demonstrated on data sets of human hands and vehicles.

Extending visual perception with haptic exploration for improved scene understanding

Scene understanding has been investigated from a mainly visual information point of view. Recently depth has been provided an extra wealth of information, allowing more geometric knowledge to fuse into scene understanding. Yet to form a holistic view, especially in robotic applications, one can create even more data by interacting with the world. In fact humans, when growing up, seem to heavily investigate the world around them by haptic exploration. We show an application of haptic exploration on a humanoid robot in cooperation with a learning method for object segmentation. The actions performed consecutively improve the segmentation of objects in the scene.

A survey on compressed domain video analysis techniques

Image and video analysis requires rich features that can characterize various aspects of visual information. These rich features are typically extracted from the pixel values of the images and videos, which require huge amount of computation and seldom useful for real-time analysis. On the contrary, the compressed domain analysis offers relevant information pertaining to the visual content in the form of transform coefficients, motion vectors, quantization steps, coded block patterns with minimal computational burden. The quantum of work done in compressed domain is relatively much less compared to pixel domain. This paper aims to survey various video analysis efforts published during the last decade across the spectrum of video compression standards. In this survey, we have included only the analysis part, excluding the processing aspect of compressed domain. This analysis spans through various computer vision applications such as moving object segmentation, human action recognition, indexing, retrieval, face detection, video classification and object tracking in compressed videos.

Computational modeling and psychophysics in low- and mid-level vision

This thesis addresses a series of topics related to the question of how people find the foreground objects from complex scenes. With both computer vision modeling, as well as psychophysical analyses, we explore the computational principles for low- and mid-level vision.

We first explore the computational methods of generating saliency maps from images and image sequences. We propose an extremely fast algorithm called Image Signature that detects the locations in the image that attract human eye gazes. With a series of experimental validations based on human behavioral data collected from various psychophysical experiments, we conclude that the Image Signature and its spatial-temporal extension, the Phase Discrepancy, are among the most accurate algorithms for saliency detection under various conditions.

In the second part, we bridge the gap between fixation prediction and salient object segmentation with two efforts. First, we propose a new dataset that contains both fixation and object segmentation information. By simultaneously presenting the two types of human data in the same dataset, we are able to analyze their intrinsic connection, as well as understanding the drawbacks of today’s “standard” but inappropriately labeled salient object segmentation dataset. Second, we also propose an algorithm of salient object segmentation. Based on our novel discoveries on the connections of fixation data and salient object segmentation data, our model significantly outperforms all existing models on all 3 datasets with large margins.

In the third part of the thesis, we discuss topics around the human factors of boundary analysis. Closely related to salient object segmentation, boundary analysis focuses on delimiting the local contours of an object. We identify the potential pitfalls of algorithm evaluation for the problem of boundary detection. Our analysis indicates that today’s popular boundary detection datasets contain significant level of noise, which may severely influence the benchmarking results. To give further insights on the labeling process, we propose a model to characterize the principles of the human factors during the labeling process.

The analyses reported in this thesis offer new perspectives to a series of interrelating issues in low- and mid-level vision. It gives warning signs to some of today’s “standard” procedures, while proposing new directions to encourage future research.