Training deep neural network policies end-to-end for real-world applications so far requires big demonstration datasets in the real world or big sets consisting of a large variety of realistic and closely related 3D CAD models. These real or virtual data should, moreover, have very similar characteristics to the conditions expected at test time. These stringent requirements and the time consuming data collection processes that they entail, are currently the most important impediment that keeps deep reinforcement learning from being deployed in real-world applications. Therefore, in this work we advocate an alternative approach, where instead of avoiding any domain shift by carefully selecting the training data, the goal is to learn a policy that can cope with it. To this end, we propose the DoShiCo challenge: to train a model in very basic synthetic environments, far from realistic, in a way that it can be applied in more realistic environments as well as take the control decisions on real-world data. In particular, we focus on the task of collision avoidance for drones. We created a set of simulated environments that can be used as benchmark and implemented a baseline method, exploiting depth prediction as an auxiliary task to help overcome the domain shift. Even though the policy is trained in very basic environments, it can learn to fly without collisions in a very different realistic simulated environment. Of course several benchmarks for reinforcement learning already exist - but they never include a large domain shift. On the other hand, several benchmarks in computer vision focus on the domain shift, but they take the form of a static datasets instead of simulated environments. In this work we claim that it is crucial to take the two challenges together in one benchmark.
Online social networks contain a constantly increasing amount of images - most of them focusing on people. Due to cultural and climate factors, fashion trends and physical appearance of individuals differ from city to city. In this paper we investigate to what extent such cues can be exploited in order to infer the geographic location, i.e. the city, where a picture was taken. We conduct a user study, as well as an evaluation of automatic methods based on convolutional neural networks. Experiments on the Fashion 144k and a Pinterest-based dataset show that the automatic methods succeed at this task to a reasonable extent. As a matter of fact, our empirical results suggest that automatic methods can surpass human performance by a large margin. Further inspection of the trained models shows that human-centered characteristics, like clothing style, physical features, and accessories, are informative for the task at hand. Moreover, it reveals that also contextual features, e.g. wall type, natural environment, etc., are taken into account by the automatic methods.
This paper proposes the novel Pose Guided Person Generation Network (PG$^2$) that allows to synthesize person images in arbitrary poses, based on an image of that person and a novel pose. Our generation framework PG$^2$ utilizes the pose information explicitly and consists of two key stages: pose integration and image refinement. In the first stage the condition image and the target pose are fed into a U-Net-like network to generate an initial but coarse image of the person with the target pose. The second stage then refines the initial and blurry result by training a U-Net-like generator in an adversarial way. Extensive experimental results on both 128$\times$64 re-identification images and 256$\times$256 fashion photos show that our model generates high-quality person images with convincing details.
Deep learning based methods have recently pushed the state-of-the-art on the problem of Single Image Super-Resolution (SISR). In this work, we revisit the more traditional interpolation-based methods, that were popular before, now with the help of deep learning. In particular, we propose to use a Convolutional Neural Network (CNN) to estimate spatially variant interpolation kernels and apply the estimated kernels adaptively to each position in the image. The whole model is trained in an end-to-end manner. We explore two ways to improve the results for the case of large upscaling factors, and propose a recursive extension of our basic model. This achieves results that are on par with state-of-the-art methods. We visualize the estimated adaptive interpolation kernels to gain more insight on the effectiveness of the proposed method. We also extend the method to the task of joint image filtering and again achieve state-of-the-art performance.
Pixelwise semantic image labeling is an important, yet challenging, task with many applications. Typical approaches to tackle this problem involve either the training of deep networks on vast amounts of images to directly infer the labels or the use of probabilistic graphical models to jointly model the dependencies of the input (i.e. images) and output (i.e. labels). Yet, the former approaches do not capture the structure of the output labels, which is crucial for the performance of dense labeling, and the latter rely on carefully hand-designed priors that require costly parameter tuning via optimization techniques, which in turn leads to long inference times. To alleviate these restrictions, we explore how to arrive at dense semantic pixel labels given both the input image and an initial estimate of the output labels. We propose a parallel architecture that: 1) exploits the context information through a LabelPropagation network to propagate correct labels from nearby pixels to improve the object boundaries, 2) uses a LabelReplacement network to directly replace possibly erroneous, initial labels with new ones, and 3) combines the different intermediate results via a Fusion network to obtain the final per-pixel label. We experimentally validate our approach on two different datasets for the semantic segmentation and face parsing tasks respectively, where we show improvements over the state-of-the-art. We also provide both a quantitative and qualitative analysis of the generated results.
This paper introduces speech-based visual question answering (VQA), the task of generating an answer given an image and a spoken question. Two methods are studied: an end-to-end, deep neural network that directly uses audio waveforms as input versus a pipelined approach that performs ASR (Automatic Speech Recognition) on the question, followed by text-based visual question answering. Furthermore, we investigate the robustness of both methods by injecting various levels of noise into the spoken question and find both methods to be tolerate noise at similar levels.
How much does a single image reveal about the environment it was taken in? In this paper, we investigate how much of that information can be retrieved from a foreground object, combined with the background (i.e. the visible part of the environment). Assuming it is not perfectly diffuse, the foreground object acts as a complexly shaped and far-from-perfect mirror. An additional challenge is that its appearance confounds the light coming from the environment with the unknown materials it is made of. We propose a learning-based approach to predict the environment from multiple reflectance maps that are computed from approximate surface normals. The proposed method allows us to jointly model the statistics of environments and material properties. We train our system from synthesized training data, but demonstrate its applicability to real-world data. Interestingly, our analysis shows that the information obtained from objects made out of multiple materials often is complementary and leads to better performance.
The task of object viewpoint estimation has been a challenge since the early days of computer vision. To estimate the viewpoint (or pose) of an object, people have mostly looked at object intrinsic features, such as shape or appearance. Surprisingly, informative features provided by other, extrinsic elements in the scene, have so far mostly been ignored. At the same time, contextual cues have been proven to be of great benefit for related tasks such as object detection or action recognition. In this paper, we explore how information from other objects in the scene can be exploited for viewpoint estimation. In particular, we look at object configurations by following a relational neighbor-based approach for reasoning about object relations. We show that, starting from noisy object detections and viewpoint estimates, exploiting the estimated viewpoint and location of other objects in the scene can lead to improved object viewpoint predictions. Experiments on the KITTI dataset demonstrate that object configurations can indeed be used as a complementary cue to appearance-based viewpoint estimation. Our analysis reveals that the proposed context-based method can improve object viewpoint estimation by reducing specific types of viewpoint estimation errors commonly made by methods that only consider local information. Moreover, considering contextual information produces superior performance in scenes where a high number of object instances occur. Finally, our results suggest that, following a cautious relational neighbor formulation brings improvements over its aggressive counterpart for the task of object viewpoint estimation.
In this paper we introduce a model of lifelong learning, based on a Network of Experts. New tasks / experts are learned and added to the model sequentially, building on what was learned before. To ensure scalability of this process,data from previous tasks cannot be stored and hence is not available when learning a new task. A critical issue in such context, not addressed in the literature so far, relates to the decision which expert to deploy at test time. We introduce a set of gating autoencoders that learn a representation for the task at hand, and, at test time, automatically forward the test sample to the relevant expert. This also brings memory efficiency as only one expert network has to be loaded into memory at any given time. Further, the autoencoders inherently capture the relatedness of one task to another, based on which the most relevant prior model to be used for training a new expert, with finetuning or learning without-forgetting, can be selected. We evaluate our method on image classification and video prediction problems.
This paper introduces a new lifelong learning solution where a single model is trained for a sequence of tasks. The main challenge that vision systems face in this context is catastrophic forgetting: as they tend to adapt to the most recently seen task, they lose performance on the tasks that were learned previously. Our method aims at preserving the knowledge of the previous tasks while learning a new one by using autoencoders. For each task, an under-complete autoencoder is learned, capturing the features that are crucial for its achievement. When a new task is presented to the system, we prevent the reconstructions of the features with these autoencoders from changing, which has the effect of preserving the information on which the previous tasks are mainly relying. At the same time, the features are given space to adjust to the most recent environment as only their projection into a low dimension submanifold is controlled. The proposed system is evaluated on image classification tasks and shows a reduction of forgetting over the state-of-the-art