Diverse image captioning models aim to learn one-to-many mappings that are innate to cross-domain datasets, such as of images and texts. Current methods for this task are based on generative latent variable models, e.g. VAEs with structured latent spaces. Yet, the amount of multimodality captured by prior work is limited to that of the paired training data -- the true diversity of the underlying generative process is not fully captured. To address this limitation, we leverage the contextual descriptions in the dataset that explain similar contexts in different visual scenes. To this end, we introduce a novel factorization of the latent space, termed context-object split, to model diversity in contextual descriptions across images and texts within the dataset. Our framework not only enables diverse captioning through context-based pseudo supervision, but extends this to images with novel objects and without paired captions in the training data. We evaluate our COS-CVAE approach on the standard COCO dataset and on the held-out COCO dataset consisting of images with novel objects, showing significant gains in accuracy and diversity.
Fashion products typically feature in compositions of a variety of styles at different clothing parts. In order to distinguish images of different fashion products, we need to extract both appearance (i.e., "how to describe") and localization (i.e.,"where to look") information, and their interactions. To this end, we propose a biologically inspired framework for image-based fashion product retrieval, which mimics the hypothesized twostream visual processing system of human brain. The proposed attentional heterogeneous bilinear network (AHBN) consists of two branches: a deep CNN branch to extract fine-grained appearance attributes and a fully convolutional branch to extract landmark localization information. A joint channel-wise attention mechanism is further applied to the extracted heterogeneous features to focus on important channels, followed by a compact bilinear pooling layer to model the interaction of the two streams. Our proposed framework achieves satisfactory performance on three image-based fashion product retrieval benchmarks.
The field of Machine Learning, a subset of Artificial Intelligence, has led to remarkable advancements in many areas, including medicine. Machine Learning algorithms require large datasets to train computer models successfully. Although there are medical image datasets available, more image datasets are needed from a variety of medical entities, especially cancer pathology. Even more scarce are ML-ready image datasets. To address this need, we created an image dataset (LC25000) with 25,000 color images in 5 classes. Each class contains 5,000 images of the following histologic entities: colon adenocarcinoma, benign colonic tissue, lung adenocarcinoma, lung squamous cell carcinoma, and benign lung tissue. All images are de-identified, HIPAA compliant, validated, and freely available for download to AI researchers.
In nuclear medicine, radioiodine therapy is prescribed to treat diseases like hyperthyroidism. The calculation of the prescribed dose depends, amongst other factors, on the thyroid volume. This is currently estimated using conventional 2D ultrasound imaging. However, this modality is inherently user-dependant, resulting in high variability in volume estimations. To increase reproducibility and consistency, we uniquely combine a neural network-based segmentation with an automatic robotic ultrasound scanning for thyroid volumetry. The robotic acquisition is achieved by using a 6 DOF robotic arm with an attached ultrasound probe. Its movement is based on an online segmentation of each thyroid lobe and the appearance of the US image. During post-processing, the US images are segmented to obtain a volume estimation. In an ablation study, we demonstrated the superiority of the motion guidance algorithms for the robot arm movement compared to a naive linear motion, executed by the robot in terms of volumetric accuracy. In a user study on a phantom, we compared conventional 2D ultrasound measurements with our robotic system. The mean volume measurement error of ultrasound expert users could be significantly decreased from 20.85+/-16.10% to only 8.23+/-3.10% compared to the ground truth. This tendency was observed even more in non-expert users where the mean error improvement with the robotic system was measured to be as high as $85\%$ which clearly shows the advantages of the robotic support.
Neurodegenerative parkinsonism can be assessed by dopamine transporter single photon emission computed tomography (DaT-SPECT). Although generating images is time-consuming, these images can show interobserver variability and they have been visually interprete by nuclear medicine physicians to date. Accordingly, this study aims to provide an automatic and robust method based on Diffusion Maps and machine learning classifiers to classify the SPECT images into two types, namely Normal and Abnormal DaT-SPECT image groups. In the proposed method, the 3D images of N patients are mapped to an N by N pairwise distance matrix and training set are embedded into a low-dimensional space by using diffusion maps. Moreover, we use Nystr\"om's out-of-sample extension, which embeds new sample points as the testing set in the reduced space. Testing samples in the embedded space are then classified into two types through the ensemble classifier with Linear Discriminant Analysis (LDA) and voting procedure through twenty-five-fold cross-validation results. The feasibility of the method is demonstrated via Parkinsonism Progression Markers Initiative (PPMI) dataset of 1097 subjects and a clinical cohort from Kaohsiung Chang Gung Memorial Hospital (KCGMH-TW) of 630 patients. We compare performances using Diffusion Maps with those of three alternative manifold methods for dimension reduction, namely Locally Linear Embedding (LLE), Isomorphic Mapping Algorithm (Isomap), and Kernel Principal Component Analysis (Kernel PCA). We also compare results using through 2D and 3D CNN methods. The diffusion maps method has an average accuracy of 98% from the PPMI and 90% from the KCGMH-TW dataset with twenty-five fold cross-validation results. It outperforms the other three methods concerning the overall accuracy and the robustness in the training and testing samples.
Single image dehazing, which aims to recover the clear image solely from an input hazy or foggy image, is a challenging ill-posed problem. Analysing existing approaches, the common key step is to estimate the haze density of each pixel. To this end, various approaches often heuristically designed haze-relevant features. Several recent works also automatically learn the features via directly exploiting Convolutional Neural Networks (CNN). However, it may be insufficient to fully capture the intrinsic attributes of hazy images. To obtain effective features for single image dehazing, this paper presents a novel Ranking Convolutional Neural Network (Ranking-CNN). In Ranking-CNN, a novel ranking layer is proposed to extend the structure of CNN so that the statistical and structural attributes of hazy images can be simultaneously captured. By training Ranking-CNN in a well-designed manner, powerful haze-relevant features can be automatically learned from massive hazy image patches. Based on these features, haze can be effectively removed by using a haze density prediction model trained through the random forest regression. Experimental results show that our approach outperforms several previous dehazing approaches on synthetic and real-world benchmark images. Comprehensive analyses are also conducted to interpret the proposed Ranking-CNN from both the theoretical and experimental aspects.
Monocular depth estimation aims at predicting depth from a single image or video. Recently, self-supervised methods draw much attention since they are free of depth annotations and achieve impressive performance on several daytime benchmarks. However, they produce weird outputs in more challenging nighttime scenarios because of low visibility and varying illuminations, which bring weak textures and break brightness-consistency assumption, respectively. To address these problems, in this paper we propose a novel framework with several improvements: (1) we introduce Priors-Based Regularization to learn distribution knowledge from unpaired depth maps and prevent model from being incorrectly trained; (2) we leverage Mapping-Consistent Image Enhancement module to enhance image visibility and contrast while maintaining brightness consistency; and (3) we present Statistics-Based Mask strategy to tune the number of removed pixels within textureless regions, using dynamic statistics. Experimental results demonstrate the effectiveness of each component. Meanwhile, our framework achieves remarkable improvements and state-of-the-art results on two nighttime datasets.
Deep neural networks (DNNs) have become ubiquitous techniques in mobile and embedded systems for applications such as image/object recognition and classification. The trend of executing multiple DNNs simultaneously exacerbate the existing limitations of meeting stringent latency/accuracy requirements on resource constrained mobile devices. The prior art sheds light on exploring the accuracy-resource tradeoff by scaling the model sizes in accordance to resource dynamics. However, such model scaling approaches face to imminent challenges: (i) large space exploration of model sizes, and (ii) prohibitively long training time for different model combinations. In this paper, we present LegoDNN, a lightweight, block-grained scaling solution for running multi-DNN workloads in mobile vision systems. LegoDNN guarantees short model training times by only extracting and training a small number of common blocks (e.g. 5 in VGG and 8 in ResNet) in a DNN. At run-time, LegoDNN optimally combines the descendant models of these blocks to maximize accuracy under specific resources and latency constraints, while reducing switching overhead via smart block-level scaling of the DNN. We implement LegoDNN in TensorFlow Lite and extensively evaluate it against state-of-the-art techniques (FLOP scaling, knowledge distillation and model compression) using a set of 12 popular DNN models. Evaluation results show that LegoDNN provides 1,296x to 279,936x more options in model sizes without increasing training time, thus achieving as much as 31.74% improvement in inference accuracy and 71.07% reduction in scaling energy consumptions.
Physics and computer science have a long tradition of cross-fertilization. One of the latest outcomes of this mutually beneficial relationship is quantum information science, which comprises the study of information processing tasks that can be accomplished using quantum mechanical systems. Quantum Image Processing (QIMP) is an emergent field of quantum information science whose main goal is to strengthen our capacity for storing, processing, and retrieving visual information from images and video either by transitioning from digital to quantum paradigms or by complementing digital imaging with quantum techniques. The expectation is that harnessing the properties of quantum mechanical systems in QIMP will result in the realization of advanced technologies that will outperform, enhance or complement existing and upcoming digital technologies for image and video processing tasks.
Research exploring CycleGAN-based synthetic image generation has recently accelerated in the medical community, as it is able to leverage unpaired datasets effectively. However, clinical acceptance of these synthetic images pose a significant challenge as they are subject to strict evaluation protocols. A commonly established drawback of the CycleGAN, the introduction of artifacts in generated images is unforgivable in the case of medical images. In an attempt to alleviate this drawback, we explore different constraints of the CycleGAN along with investigation of adaptive control of these constraints. The benefits of imposing additional constraints on the CycleGAN, in the form of structure retaining losses is also explored. A generalized frequency loss inspired by arxiv:2012.12821 that preserves content in the frequency domain between source and target is investigated and compared with existing losses such as the MIND loss arXiv:1809.04536. CycleGAN implementations from the ganslate framework (https://github.com/ganslate-team/ganslate) are used for experimentation in this thesis. Synthetic images generated from our methods are quantitatively and qualitatively investigated and outperform the baseline CycleGAN and other approaches. Furthermore, no observable artifacts or loss in image quality is found, which is critical for acceptance of these synthetic images. The synthetic medical images thus generated are also evaluated using domain-specific evaluation and using segmentation as a downstream task, in order to clearly highlight their applicability to clinical workflows.