Abstract: In this article, we focus on the problem of large-scale instance-level image retrieval. For efficiency reasons, it is common to represent an image by a fixed-length descriptor which is subsequently encoded into a small number of bits. We note that most encoding techniques include an unsupervised dimensionality reduction step. Our goal in this work is to learn a better subspace in a supervised manner. We especially raise the following question: “can category-level labels be used to learn such a subspace?” To answer this question, we experiment with four learning techniques: the first one is based on a metric learning framework, the second one on attribute representations, the third one on Canonical Correlation Analysis (CCA) and the fourth one on Joint Subspace and Classifier Learning (JSCL). While the first three approaches have been applied in the past to the image retrieval problem, we believe we are the first to show the usefulness of JSCL in this context. In our experiments, we use ImageNet as a source of category-level labels and report retrieval results on two standard dataseis: INRIA Holidays and the University of Kentucky benchmark. Our experimental study shows that metric learning and attributes do not lead to any significant improvement in retrieval accuracy, as opposed to CCA and JSCL. As an example, we report on Holidays an increase in accuracy from 39.3% to 48.6% with 32-dimensional representations. Overall JSCL is shown to yield the best results.

Abstract: The study of neural generative models of handwritten text and human sketches is a hot topic in the computer vision field. The landmark SketchRNN provided a breakthrough by sequentially generating sketches as a sequence of waypoints, and more recent articles have managed to generate fully vector sketches by coding the strokes as Bézier curves. However, the previous attempts with this approach need them all a ground truth consisting in the sequence of points that make up each stroke, which seriously limits the datasets the model is able to train in. In this work, we present a self-supervised end-to-end inverse graphics approach that learns to embed each image to its best fit of Bézier curves. The self-supervised nature of the training process allows us to train the model in a wider range of datasets, but also to perform better after-training predictions by applying an overfitting process on the input binary image. We report qualitative an quantitative evaluations on the MNIST and the Quick, Draw! datasets.

Abstract: Vision and Language are broadly regarded as cornerstones of intelligence. Even though language and vision have different aims – language having the purpose of communication, transmission of information and vision having the purpose of constructing mental representations around us to navigate and interact with objects – they cooperate and depend on one another in many tasks we perform effortlessly. This reliance is actively being studied in various Computer Vision tasks, e.g. image captioning, visual question answering, image-sentence retrieval, phrase grounding, just to name a few. All of these tasks share the inherent difficulty of the aligning the two modalities, while being robust to language
priors and various biases existing in the datasets. One of the ultimate goal for vision and language research is to be able to inject world knowledge while getting rid of the biases that come with the datasets. In this thesis, we mainly focus on two vision and language tasks, namely Image Captioning and Scene-Text Visual Question Answering (STVQA).
In both domains, we start by defining a new task that requires the utilization of world knowledge and in both tasks, we find that the models commonly employed are prone to biases that exist in the data. Concretely, we introduce new tasks and discover several problems that impede performance at each level and provide remedies or possible solutions in each chapter: i) We define a new task to move beyond Image Captioning to Image Interpretation that can utilize Named Entities in the form of world knowledge. ii) We study the object hallucination problem in classic Image Captioning systems and develop an architecture-agnostic solution. iii) We define a sub-task of Visual Question Answering that requires reading the text in the image (STVQA), where we highlight the limitations of current models. iv) We propose an architecture for the STVQA task that can point to the answer in the image and show how to combine it with classic VQA models. v) We show how far language can get us in STVQA and discover yet another bias which causes the models to disregard the image while doing Visual Question Answering.

Abstract: Explaining an image with missing or non-existent objects is known as object bias (hallucination) in image captioning. This behaviour is quite common in the state-of-the-art captioning models which is not desirable by humans. To decrease the object hallucination in captioning, we propose three simple yet efficient training augmentation method for sentences which requires no new training data or increase
in the model size. By extensive analysis, we show that the proposed methods can significantly diminish our models’ object bias on hallucination metrics. Moreover, we experimentally demonstrate that our methods decrease the dependency on the visual features. All of our code, configuration files and model weights are available online.

Abstract: Current visual question answering datasets do not consider the rich semantic information conveyed by text within an image. In this work, we present a new dataset, ST-VQA, that aims to highlight the importance of exploiting highlevel semantic information present in images as textual cues in the Visual Question Answering process. We use this dataset to define a series of tasks of increasing difficulty for which reading the scene text in the context provided by the visual information is necessary to reason and generate an appropriate answer. We propose a new evaluation metric for these tasks to account both for reasoning errors as well as shortcomings of the text recognition module. In addition we put forward a series of baseline methods, which provide further insight to the newly released dataset, and set the scene for further research.