arXiv subjects

Tomer Peleg

Publications and source records attributed to Tomer Peleg.

6 recordsLinked to original sources

arXiv2024

Do More With What You Have: Transferring Depth-Scale from Labeled to Unlabeled Domains

Transferring the absolute depth prediction capabilities of an estimator to a new domain is a task with significant real-world applications. This task is specifically challenging when images from the new domain are collected without ground-truth depth measurements, and possibly with sensors of different intrinsics. To overcome such limitations, a recent zero-shot solution was trained on an extensive training dataset and encoded the various camera intrinsics. Other solutions generated synthetic data with depth labels that matched the intrinsics of the new target data to enable depth-scale transfer between the domains. In this work we present an alternative solution that can utilize any existing synthetic or real dataset, that has a small number of images annotated with ground truth depth labels. Specifically, we show that self-supervised depth estimators result in up-to-scale predictions that are linearly correlated to their absolute depth values across the domain, a property that we model in this work using a single scalar. In addition, aligning the field-of-view of two datasets prior to training, results in a common linear relationship for both domains. We use this observed property to transfer the depth-scale from source datasets that have absolute depth labels to new target datasets that lack these measurements, enabling absolute depth predictions in the target domain. The suggested method was successfully demonstrated on the KITTI, DDAD and nuScenes datasets, while using other existing real or synthetic source datasets, that have a different field-of-view, other image style or structural content, achieving comparable or better accuracy than other existing methods that do not use target ground-truth depths.

cs.CV↗

arXiv2024

SDAT: Sub-Dataset Alternation Training for Improved Image Demosaicing

Image demosaicing is an important step in the image processing pipeline for digital cameras. In data centric approaches, such as deep learning, the distribution of the dataset used for training can impose a bias on the networks' outcome. For example, in natural images most patches are smooth, and high-content patches are much rarer. This can lead to a bias in the performance of demosaicing algorithms. Most deep learning approaches address this challenge by utilizing specific losses or designing special network architectures. We propose a novel approach, SDAT, Sub-Dataset Alternation Training, that tackles the problem from a training protocol perspective. SDAT is comprised of two essential phases. In the initial phase, we employ a method to create sub-datasets from the entire dataset, each inducing a distinct bias. The subsequent phase involves an alternating training process, which uses the derived sub-datasets in addition to training also on the entire dataset. SDAT can be applied regardless of the chosen architecture as demonstrated by various experiments we conducted for the demosaicing task. The experiments are performed across a range of architecture sizes and types, namely CNNs and transformers. We show improved performance in all cases. We are also able to achieve state-of-the-art results on three highly popular image demosaicing benchmarks.

eess.IV↗

arXiv2021

You Better Look Twice: a new perspective for designing accurate detectors with reduced computations

General object detectors use powerful backbones that uniformly extract features from images for enabling detection of a vast amount of object types. However, utilization of such backbones in object detection applications developed for specific object types can unnecessarily over-process an extensive amount of background. In addition, they are agnostic to object scales, thus redundantly process all image regions at the same resolution. In this work we introduce BLT-net, a new low-computation two-stage object detection architecture designed to process images with a significant amount of background and objects of variate scales. BLT-net reduces computations by separating objects from background using a very lite first-stage. BLT-net then efficiently merges obtained proposals to further decrease processed background and then dynamically reduces their resolution to minimize computations. Resulting image proposals are then processed in the second-stage by a highly accurate model. We demonstrate our architecture on the pedestrian detection problem, where objects are of different sizes, images are of high resolution and object detection is required to run in real-time. We show that our design reduces computations by a factor of x4-x7 on the Citypersons and Caltech datasets with respect to leading pedestrian detectors, on account of a small accuracy degradation. This method can be applied on other object detection applications in scenes with a considerable amount of background and variate object sizes to reduce computations.

cs.CV↗

arXiv2014

Fundamental performance limits for ideal decoders in high-dimensional linear inverse problems

This paper focuses on characterizing the fundamental performance limits that can be expected from an ideal decoder given a general model, ie, a general subset of "simple" vectors of interest. First, we extend the so-called notion of instance optimality of a decoder to settings where one only wishes to reconstruct some part of the original high dimensional vector from a low-dimensional observation. This covers practical settings such as medical imaging of a region of interest, or audio source separation when one is only interested in estimating the contribution of a specific instrument to a musical recording. We define instance optimality relatively to a model much beyond the traditional framework of sparse recovery, and characterize the existence of an instance optimal decoder in terms of joint properties of the model and the considered linear operator. Noiseless and noise-robust settings are both considered. We show somewhat surprisingly that the existence of noise-aware instance optimal decoders for all noise levels implies the existence of a noise-blind decoder. A consequence of our results is that for models that are rich enough to contain an orthonormal basis, the existence of an L2/L2 instance optimal decoder is only possible when the linear operator is not substantially dimension-reducing. This covers well-known cases (sparse vectors, low-rank matrices) as well as a number of seemingly new situations (structured sparsity and sparse inverse covariance matrices for instance). We exhibit an operator-dependent norm which, under a model-specific generalization of the Restricted Isometry Property (RIP), always yields a feasible instance optimality property. This norm can be upper bounded by an atomic norm relative to the considered model.

cs.IT↗

arXiv2012

Exploiting Statistical Dependencies in Sparse Representations for Signal Recovery

Signal modeling lies at the core of numerous signal and image processing applications. A recent approach that has drawn considerable attention is sparse representation modeling, in which the signal is assumed to be generated as a combination of a few atoms from a given dictionary. In this work we consider a Bayesian setting and go beyond the classic assumption of independence between the atoms. The main goal of this paper is to introduce a statistical model that takes such dependencies into account and show how this model can be used for sparse signal recovery. We follow the suggestion of two recent works and assume that the sparsity pattern is modeled by a Boltzmann machine, a commonly used graphical model. For general dependency models, exact MAP and MMSE estimation of the sparse representation becomes computationally complex. To simplify the computations, we propose greedy approximations of the MAP and MMSE estimators. We then consider a special case in which exact MAP is feasible, by assuming that the dictionary is unitary and the dependency model corresponds to a certain sparse graph. Exploiting this structure, we develop an efficient message passing algorithm that recovers the underlying signal. When the model parameters defining the underlying graph are unknown, we suggest an algorithm that learns these parameters directly from the data, leading to an iterative scheme for adaptive sparse signal recovery. The effectiveness of our approach is demonstrated on real-life signals - patches of natural images - where we compare the denoising performance to that of previous recovery methods that do not exploit the statistical dependencies.

cs.IT↗

arXiv2012

Performance Guarantees of the Thresholding Algorithm for the Co-Sparse Analysis Model

The co-sparse analysis model for signals assumes that the signal of interest can be multiplied by an analysis dictionary Ω, leading to a sparse outcome. This model stands as an interesting alternative to the more classical synthesis based sparse representation model. In this work we propose a theoretical study of the performance guarantee of the thresholding algorithm for the pursuit problem in the presence of noise. Our analysis reveals two significant properties of Ω, which govern the pursuit performance: The first is the degree of linear dependencies between sets of rows in Ω, depicted by the co-sparsity level. The second property, termed the Restricted Orthogonal Projection Property (ROPP), is the level of independence between such dependent sets and other rows in Ω. We show how these dictionary properties are meaningful and useful, both in the theoretical bounds derived, and in a series of experiments that are shown to align well with the theoretical prediction.

cs.IT↗