In this paper we propose an unsupervised feature extraction method to capture temporal information on monocular videos, where we detect and encode subject of interest in each frame and leverage contrastive self-supervised (CSS) learning to extract rich latent vectors. Instead of simply treating the latent features of nearby frames as positive pairs and those of temporally-distant ones as negative pairs as in other CSS approaches, we explicitly disentangle each latent vector into a time-variant component and a time-invariant one. We then show that applying contrastive loss only to the time-variant features and encouraging a gradual transition on them between nearby and away frames while also reconstructing the input, extract rich temporal features, well-suited for human pose estimation. Our approach reduces error by about 50% compared to the standard CSS strategies, outperforms other unsupervised single-view methods and matches the performance of multi-view techniques. When 2D pose is available, our approach can extract even richer latent features and improve the 3D pose estimation accuracy, outperforming other state-of-the-art weakly supervised methods.

In this paper, we tackle the task of estimating the 3D orientation of previously-unseen objects from monocular images. This task contrasts with the one considered by most existing deep learning methods which typically assume that the testing objects have been observed during training. To handle the unseen objects, we follow a retrieval-based strategy and prevent the network from learning object-specific features by computing multi-scale local similarities between the query image and synthetically-generated reference images. We then introduce an adaptive fusion module that robustly aggregates the local similarities into a global similarity score of pairwise images. Furthermore, we speed up the retrieval process by developing a fast retrieval strategy. Our experiments on the LineMOD, LineMOD-Occluded, and T-LESS datasets show that our method yields a significantly better generalization to unseen objects than previous works. Our code and pre-trained models are available at https://sailor-z.github.io/projects/Unseen_Object_Pose.html.

Most recent 6D object pose estimation methods, including unsupervised ones, require many real training images. Unfortunately, for some applications, such as those in space or deep under water, acquiring real images, even unannotated, is virtually impossible. In this paper, we propose a method that can be trained solely on synthetic images, or optionally using a few additional real ones. Given a rough pose estimate obtained from a first network, it uses a second network to predict a dense 2D correspondence field between the image rendered using the rough pose and the real image and infers the required pose correction. This approach is much less sensitive to the domain shift between synthetic and real images than state-of-the-art methods. It performs on par with methods that require annotated real images for training when not using any, and outperforms them considerably when using as few as twenty real images.

Knowledge distillation constitutes a simple yet effective way to improve the performance of a compact student network by exploiting the knowledge of a more powerful teacher. Nevertheless, the knowledge distillation literature remains limited to the scenario where the student and the teacher tackle the same task. Here, we investigate the problem of transferring knowledge not only across architectures but also across tasks. To this end, we study the case of object detection and, instead of following the standard detector-to-detector distillation approach, introduce a classifier-to-detector knowledge transfer framework. In particular, we propose strategies to exploit the classification teacher to improve both the detector's recognition accuracy and localization performance. Our experiments on several detectors with different backbones demonstrate the effectiveness of our approach, allowing us to outperform the state-of-the-art detector-to-detector distillation methods.

We analyze the influence of adversarial training on the loss landscape of machine learning models. To this end, we first provide analytical studies of the properties of adversarial loss functions under different adversarial budgets. We then demonstrate that the adversarial loss landscape is less favorable to optimization, due to increased curvature and more scattered gradients. Our conclusions are validated by numerical analyses, which show that training under large adversarial budgets impede the escape from suboptimal random initialization, cause non-vanishing gradients and make the model find sharper minima. Based on these observations, we show that a periodic adversarial scheduling (PAS) strategy can effectively overcome these challenges, yielding better results than vanilla adversarial training while being much less sensitive to the choice of learning rate.

Recent approaches to multi-view learning have shown that factorizing the information into parts that are shared across all views and parts that are private to each view could effectively account for the dependencies and independencies between the different input modalities. Unfortunately, these approaches involve minimizing non-convex objective functions. In this paper, we propose an approach to learning such factorized representations inspired by sparse coding techniques. In particular, we show that structured sparsity allows us to address the multi-view learning problem by alternately solving two convex optimization problems. Furthermore, the resulting factorized latent spaces generalize over existing approaches in that they allow:having latent dimensions shared between any subset of the views instead of between all the views only.

Human motion prediction is a stochastic process: Given an observed sequence of poses, multiple future motions are plausible. Existing approaches to modeling this stochasticity typically combine a random noise vector with information about the previous poses. This combination, however, is done in a deterministic manner, which gives the network the flexibility to learn to ignore the random noise. In this paper, we introduce an approach to stochastically combine the root of variations with previous pose information, which forces the model to take the noise into account. W e exploit this idea for motion prediction by incorporating it into a recurrent encoder-decoder network with a conditional vari-ational autoencoder block that learns to exploit the perturbations. Our experiments demonstrate that our model yields high-quality pose sequences that are much more diverse than those from state-of-the-art stochastic motion prediction techniques.

Eigendecomposition (ED) is widely used in deep networks. However, the backpropagation of its results tends to be numerically unstable, whether using ED directly or approximating it with the Power Iteration method, particularly when dealing with large matrices. While this can be mitigated by partitioning the data in small and arbitrary groups, doing so has no theoretical basis and makes its impossible to exploit the power of ED to the full. In this paper, we introduce a numerically stable and differentiable approach to leveraging eigenvectors in deep networks. It can handle large matrices without requiring to split them. We demonstrate the better robustness of our approach over standard ED and PI for ZCA whitening, an alternative to batch normalization, and for PCA denoising, which we introduce as a new normalization strategy for deep networks, aiming to further denoise the network's features.

When We identify a phenomenon, which we refer to dealing with many large models, a common strategy to keep as multi-model forgetting, that occurs when sequentially training tractable is to share a subset of the weights across training multiple deep networks with the multiple models and to train them sequentially (Pham partially-shared parameters; the performance of et al., 2018; Xie & Yuille, 2017; Liu et al., 2018a). This previously-trained models degrades as one optimizes strategy has a major drawback. Figure 1 shows that for two a subsequent one, due to the overwriting models, A and B, the larger the number of shared weights, of shared parameters. To overcome this, we introduce the more the accuracy of A drops when training B; B overwrites a statistically-justified weight plasticity loss some of the weights of A and this damages the performance that regularizes the learning of a model's shared of A. We call this multi-model forgetting. The parameters according to their importance for the benefits of weight-sharing have been emphasized in tasks previous models, and demonstrate its effectiveness like neural architecture search, where the associated speed when training two models sequentially and gains have been key in making the process practical (Pham for neural architecture search. Adding weight et al., 2018; Liu et al., 2018b), but its downsides remain plasticity in neural architecture search preserves virtually unexplored.

Neural Architecture Search (NAS) aims to facilitate the design of deep networks for new tasks. Existing techniques rely on two stages: searching over the architecture space and validating the best architecture. Evaluating NAS algorithms is currently solely done by comparing their results on the downstream task. While intuitive, this fails to explicitly evaluate the effectiveness of their search strategies. In this paper, we extend the NAS evaluation procedure to include the search phase. To this end, we compare the quality of the solutions obtained by NAS search policies with that of random architecture selection. We find that: (i) On average, the random policy outperforms state-of-the-art NAS algorithms; and (ii) The results and candidate rankings of NAS algorithms do not reflect the true performance of the candidate architectures. While our former finding illustrates the fact that the NAS search space has been sufficiently constrained so that random solutions yield good results, we trace the latter back to the weight sharing strategy used by state-of-the-art NAS methods. In contrast with common belief, weight sharing negatively impacts the training of good architectures, thus reducing the effectiveness of the search process. We believe that following our evaluation framework will be key to designing NAS strategies that truly discover superior architectures.