In this paper, we propose a new object detection model, Mixture-Model-based Object Detector (MMOD), that performs multi-object detection using a mixture model. Unlike previous studies, we use density estimation to deal with the multi-object detection task. MMOD captures the conditional distribution of bounding boxes for a given input image using a mixture model consisting of Gaussian and categorical distributions. For this purpose, we propose a method to extract object bounding boxes from a trained mixture model. In doing so, we also propose a new network structure and objective function for the MMOD. Our proposed method is not trained by assigning a ground truth bounding box to a specific location on the network's output. Instead, the mixture components are automatically learned to represent the distribution of the bounding box through density estimation. Therefore, MMOD does not require a large number of anchors and does not incur the positive-negative imbalance problem. This not only benefits the detection performance but also enhances the inference speed without requiring additional processing. We applied MMOD to Pascal VOC and MS COCO datasets, and outperform the detection performance with inference speed of other state-of-the-art fast object detection methods. (38.7 AP with 39ms per image on MS COCO without bells and whistles.) Code will be available.
Person search is to detect all persons and identify the query persons from detected persons in the image without proposals and bounding boxes, which is different from person re-identification. In this paper, we propose a fusing multi-task convolutional neural network(FMT-CNN) to tackle the correlation and heterogeneity of detection and re-identification with a single convolutional neural network. We focus on how the interplay of person detection and person re-identification affects the overall performance. We employ person labels in region proposal network to produce features for person re-identification and person detection network, which can improve the accuracy of detection and re-identification simultaneously. We also use a multiple loss to train our re-identification network. Experiment results on CUHK-SYSU Person Search dataset show that the performance of our proposed method is superior to state-of-the-art approaches in both mAP and top-1.
Sparse representation with respect to an overcomplete dictionary is often used when regularizing inverse problems in signal and image processing. In recent years, the Convolutional Sparse Coding (CSC) model, in which the dictionary consists of shift-invariant filters, has gained renewed interest. While this model has been successfully used in some image processing problems, it still falls behind traditional patch-based methods on simple tasks such as denoising. In this work we provide new insights regarding the CSC model and its capability to represent natural images, and suggest a Bayesian connection between this model and its patch-based ancestor. Armed with these observations, we suggest a novel feed-forward network that follows an MMSE approximation process to the CSC model, using strided convolutions. The performance of this supervised architecture is shown to be on par with state of the art methods while using much fewer parameters.
Magnetic Resonance Imaging (MRI) is a non-invasive diagnostic tool that provides excellent soft-tissue contrast without the use of ionizing radiation. But, compared to other clinical imaging modalities (e.g., CT or ultrasound), the data acquisition process for MRI is inherently slow. Furthermore, dynamic applications demand collecting a series of images in quick succession. As a result, reducing acquisition time and improving imaging quality for undersampled datasets have been active areas of research for the last two decades. The combination of parallel imaging and compressive sensing (CS) has been shown to benefit a wide range of MRI applications. More recently, deep learning techniques have been shown to outperform CS methods. Some of these techniques pose the MRI reconstruction as a direct inversion problem and tackle it by training a deep neural network (DNN) to map from the measured Fourier samples and the final image. Considering that the forward model in MRI changes from one dataset to the next, such methods have to be either trained over a large and diverse corpus of data or limited to a specific application, and even then they cannot ensure data consistency. An alternative is to use "plug-and-play" (PnP) algorithms, which iterate image denoising with forward-model based signal recovery. PnP algorithms are an excellent fit for compressive MRI because they decouple image modeling from the forward model, which can change significantly among different scans due to variations in the coil sensitivity maps, sampling patterns, and image resolution. Consequently, with PnP, state-of-the-art image-denoising techniques, such as those based on DNNs, can be directly exploited for compressive MRI image reconstruction. The objective of this article is two-fold: i) to review recent advances in plug-and-play methods, and ii) to discuss their application to compressive MRI image reconstruction.
There has long been debates on how we could interpret neural networks and understand the decisions our models make. Specifically, why deep neural networks tend to be error-prone when dealing with samples that output low softmax scores. We present an efficient approach to measure the confidence of decision-making steps by statistically investigating each unit's contribution to that decision. Instead of focusing on how the models react on datasets, we study the datasets themselves given a pre-trained model. Our approach is capable of assigning a score to each sample within a dataset that measures the frequency of occurrence of that sample's chain of activation. We demonstrate with experiments that our method could select useful samples to improve deep neural networks in a semi-supervised leaning setting.
Predictive marker patterns in imaging data are a means to quantify disease and progression, but their identification is challenging, if the underlying biology is poorly understood. Here, we present a method to identify predictive texture patterns in medical images in an unsupervised way. Based on deep clustering networks, we simultaneously encode and cluster medical image patches in a low-dimensional latent space. The resulting clusters serve as features for disease staging, linking them to the underlying disease. We evaluate the method on 70 T1-weighted magnetic resonance images of patients with different stages of liver steatosis. The deep clustering approach is able to find predictive clusters with a stable ranking, differentiating between low and high steatosis with an F1-Score of 0.78.
Panoramic images have advantages in information capacity and scene stability due to their large field of view (FoV). In this paper, we propose a method to synthesize a new dataset of panoramic image. We managed to stitch the images taken from different directions into panoramic images, together with their labeled images, to yield the panoramic semantic segmentation dataset denominated as SYNTHIA-PANO. For the purpose of finding out the effect of using panoramic images as training dataset, we designed and performed a comprehensive set of experiments. Experimental results show that using panoramic images as training data is beneficial to the segmentation result. In addition, it has been shown that by using panoramic images with a 180 degree FoV as training data the model has better performance. Furthermore, the model trained with panoramic images also has a better capacity to resist the image distortion.
Neural Architecture Search (NAS) has achieved great success in image classification task. Some recent works have managed to explore the automatic design of efficient backbone or feature fusion layer for object detection. However, these methods focus on searching only one certain component of object detector while leaving others manually designed. We identify the inconsistency between searched component and manually designed ones would withhold the detector of stronger performance. To this end, we propose a hierarchical trinity search framework to simultaneously discover efficient architectures for all components (i.e. backbone, neck, and head) of object detector in an end-to-end manner. In addition, we empirically reveal that different parts of the detector prefer different operators. Motivated by this, we employ a novel scheme to automatically screen different sub search spaces for different components so as to perform the end-to-end search for each component on the corresponding sub search space efficiently. Without bells and whistles, our searched architecture, namely Hit-Detector, achieves 41.4\% mAP on COCO minival set with 27M parameters. Our implementation is available at https://github.com/ggjy/HitDet.pytorch.
Image restoration is one of the most fundamental issues in imaging science. Total variation (TV) regularization is widely used in image restoration problems for its capability to preserve edges. In the literature, however, it is also well known for producing staircase-like artifacts. Usually, the high-order total variation (HTV) regularizer is an good option except its over-smoothing property. In this work, we study a minimization problem where the objective includes an usual $l_2$ data-fidelity term and an overlapping group sparsity total variation regularizer which can avoid staircase effect and allow edges preserving in the restored image. We also proposed a fast algorithm for solving the corresponding minimization problem and compare our method with the state-of-the-art TV based methods and HTV based method. The numerical experiments illustrate the efficiency and effectiveness of the proposed method in terms of PSNR, relative error and computing time.
By developing sophisticated image priors or designing deep(er) architectures, a variety of image Super-Resolution (SR) approaches have been proposed recently and achieved very promising performance. A natural question that arises is whether these methods can be reformulated into a unifying framework and whether this framework assists in SR reconstruction? In this paper, we present a simple but effective single image SR method based on ensemble learning, which can produce a better performance than that could be obtained from any of SR methods to be ensembled (or called component super-resolvers). Based on the assumption that better component super-resolver should have larger ensemble weight when performing SR reconstruction, we present a Maximum A Posteriori (MAP) estimation framework for the inference of optimal ensemble weights. Specially, we introduce a reference dataset, which is composed of High-Resolution (HR) and Low-Resolution (LR) image pairs, to measure the super-resolution abilities (prior knowledge) of different component super-resolvers. To obtain the optimal ensemble weights, we propose to incorporate the reconstruction constraint, which states that the degenerated HR image should be equal to the LR observation one, as well as the prior knowledge of ensemble weights into the MAP estimation framework. Moreover, the proposed optimization problem can be solved by an analytical solution. We study the performance of the proposed method by comparing with different competitive approaches, including four state-of-the-art non-deep learning based methods, four latest deep learning based methods and one ensemble learning based method, and prove its effectiveness and superiority on three public datasets.