arXiv subjects

Ariel Shamir

Publications and source records attributed to Ariel Shamir.

68 records · Page 4Linked to original sources

arXiv2022

CAST: Character labeling in Animation using Self-supervision by Tracking

Cartoons and animation domain videos have very different characteristics compared to real-life images and videos. In addition, this domain carries a large variability in styles. Current computer vision and deep-learning solutions often fail on animated content because they were trained on natural images. In this paper we present a method to refine a semantic representation suitable for specific animated content. We first train a neural network on a large-scale set of animation videos and use the mapping to deep features as an embedding space. Next, we use self-supervision to refine the representation for any specific animation style by gathering many examples of animated characters in this style, using a multi-object tracking. These examples are used to define triplets for contrastive loss training. The refined semantic space allows better clustering of animated characters even when they have diverse manifestations. Using this space we can build dictionaries of characters in an animation videos, and define specialized classifiers for specific stylistic content (e.g., characters in a specific animation series) with very little user effort. These classifiers are the basis for automatically labeling characters in animation videos. We present results on a collection of characters in a variety of animation styles.

cs.CV↗

arXiv2021

Rhythm is a Dancer: Music-Driven Motion Synthesis with Global Structure

Synthesizing human motion with a global structure, such as a choreography, is a challenging task. Existing methods tend to concentrate on local smooth pose transitions and neglect the global context or the theme of the motion. In this work, we present a music-driven motion synthesis framework that generates long-term sequences of human motions which are synchronized with the input beats, and jointly form a global structure that respects a specific dance genre. In addition, our framework enables generation of diverse motions that are controlled by the content of the music, and not only by the beat. Our music-driven dance synthesis framework is a hierarchical system that consists of three levels: pose, motif, and choreography. The pose level consists of an LSTM component that generates temporally coherent sequences of poses. The motif level guides sets of consecutive poses to form a movement that belongs to a specific distribution using a novel motion perceptual-loss. And the choreography level selects the order of the performed movements and drives the system to follow the global structure of a dance genre. Our results demonstrate the effectiveness of our music-driven framework to generate natural and consistent movements on various dance types, having control over the content of the synthesized motions, and respecting the overall structure of the dance.

cs.GR↗

arXiv2021

Image Resizing by Reconstruction from Deep Features

Traditional image resizing methods usually work in pixel space and use various saliency measures. The challenge is to adjust the image shape while trying to preserve important content. In this paper we perform image resizing in feature space where the deep layers of a neural network contain rich important semantic information. We directly adjust the image feature maps, extracted from a pre-trained classification network, and reconstruct the resized image using a neural-network based optimization. This novel approach leverages the hierarchical encoding of the network, and in particular, the high-level discriminative power of its deeper layers, that recognizes semantic objects and regions and allows maintaining their aspect ratio. Our use of reconstruction from deep features diminishes the artifacts introduced by image-space resizing operators. We evaluate our method on benchmarks, compare to alternative approaches, and demonstrate its strength on challenging images.

cs.CV↗

arXiv2021

InAugment: Improving Classifiers via Internal Augmentation

Image augmentation techniques apply transformation functions such as rotation, shearing, or color distortion on an input image. These augmentations were proven useful in improving neural networks' generalization ability. In this paper, we present a novel augmentation operation, InAugment, that exploits image internal statistics. The key idea is to copy patches from the image itself, apply augmentation operations on them, and paste them back at random positions on the same image. This method is simple and easy to implement and can be incorporated with existing augmentation techniques. We test InAugment on two popular datasets -- CIFAR and ImageNet. We show improvement over state-of-the-art augmentation techniques. Incorporating InAugment with Auto Augment yields a significant improvement over other augmentation techniques (e.g., +1% improvement over multiple architectures trained on the CIFAR dataset). We also demonstrate an increase for ResNet50 and EfficientNet-B3 top-1's accuracy on the ImageNet dataset compared to prior augmentation methods. Finally, our experiments suggest that training convolutional neural network using InAugment not only improves the model's accuracy and confidence but its performance on out-of-distribution images.

cs.CV↗

arXiv2020

Neural Alignment for Face De-pixelization

We present a simple method to reconstruct a high-resolution video from a face-video, where the identity of a person is obscured by pixelization. This concealment method is popular because the viewer can still perceive a human face figure and the overall head motion. However, we show in our experiments that a fairly good approximation of the original video can be reconstructed in a way that compromises anonymity. Our system exploits the simultaneous similarity and small disparity between close-by video frames depicting a human face, and employs a spatial transformation component that learns the alignment between the pixelated frames. Each frame, supported by its aligned surrounding frames, is first encoded, then decoded to a higher resolution. Reconstruction and perceptual losses promote adherence to the ground-truth, and an adversarial loss assists in maintaining domain faithfulness. There is no need for explicit temporal coherency loss as it is maintained implicitly by the alignment of neighboring frames and reconstruction. Although simple, our framework synthesizes high-quality face reconstructions, demonstrating that given the statistical prior of a human face, multiple aligned pixelated frames contain sufficient information to reconstruct a high-quality approximation of the original signal.

cs.CV↗

arXiv2020

NPRportrait 1.0: A Three-Level Benchmark for Non-Photorealistic Rendering of Portraits

Despite the recent upsurge of activity in image-based non-photorealistic rendering (NPR), and in particular portrait image stylisation, due to the advent of neural style transfer, the state of performance evaluation in this field is limited, especially compared to the norms in the computer vision and machine learning communities. Unfortunately, the task of evaluating image stylisation is thus far not well defined, since it involves subjective, perceptual and aesthetic aspects. To make progress towards a solution, this paper proposes a new structured, three level, benchmark dataset for the evaluation of stylised portrait images. Rigorous criteria were used for its construction, and its consistency was validated by user studies. Moreover, a new methodology has been developed for evaluating portrait stylisation algorithms, which makes use of the different benchmark levels as well as annotations provided by user studies regarding the characteristics of the faces. We perform evaluation for a wide variety of image stylisation methods (both portrait-specific and general purpose, and also both traditional NPR approaches and neural style transfer) using the new benchmark dataset.

cs.CV↗

arXiv2020

Focus-and-Expand: Training Guidance Through Gradual Manipulation of Input Features

We present a simple and intuitive Focus-and-eXpand (\fax) method to guide the training process of a neural network towards a specific solution. Optimizing a neural network is a highly non-convex problem. Typically, the space of solutions is large, with numerous possible local minima, where reaching a specific minimum depends on many factors. In many cases, however, a solution which considers specific aspects, or features, of the input is desired. For example, in the presence of bias, a solution that disregards the biased feature is a more robust and accurate one. Drawing inspiration from Parameter Continuation methods, we propose steering the training process to consider specific features in the input more than others, through gradual shifts in the input domain. \fax extracts a subset of features from each input data-point, and exposes the learner to these features first, Focusing the solution on them. Then, by using a blending/mixing parameter $α$ it gradually eXpands the learning process to include all features of the input. This process encourages the consideration of the desired features more than others. Though not restricted to this field, we quantitatively evaluate the effectiveness of our approach on various Computer Vision tasks, and achieve state-of-the-art bias removal, improvements to an established augmentation method, and two examples of improvements to image classification tasks. Through these few examples we demonstrate the impact this approach potentially carries for a wide variety of problems, which stand to gain from understanding the solution landscape.

cs.LG↗

arXiv2020

Predictive and Generative Neural Networks for Object Functionality

Humans can predict the functionality of an object even without any surroundings, since their knowledge and experience would allow them to "hallucinate" the interaction or usage scenarios involving the object. We develop predictive and generative deep convolutional neural networks to replicate this feat. Specifically, our work focuses on functionalities of man-made 3D objects characterized by human-object or object-object interactions. Our networks are trained on a database of scene contexts, called interaction contexts, each consisting of a central object and one or more surrounding objects, that represent object functionalities. Given a 3D object in isolation, our functional similarity network (fSIM-NET), a variation of the triplet network, is trained to predict the functionality of the object by inferring functionality-revealing interaction contexts. fSIM-NET is complemented by a generative network (iGEN-NET) and a segmentation network (iSEG-NET). iGEN-NET takes a single voxelized 3D object with a functionality label and synthesizes a voxelized surround, i.e., the interaction context which visually demonstrates the corresponding functionality. iSEG-NET further separates the interacting objects into different groups according to their interaction types.

cs.CV↗

arXiv2019

Deep Portrait Image Completion and Extrapolation

General image completion and extrapolation methods often fail on portrait images where parts of the human body need to be recovered - a task that requires accurate human body structure and appearance synthesis. We present a two-stage deep learning framework for tacking this problem. In the first stage, given a portrait image with an incomplete human body, we extract a complete, coherent human body structure through a human parsing network, which focuses on structure recovery inside the unknown region with the help of pose estimation. In the second stage, we use an image completion network to fill the unknown region, guided by the structure map recovered in the first stage. For realistic synthesis the completion network is trained with both perceptual loss and conditional adversarial loss. We evaluate our method on public portrait image datasets, and show that it outperforms other state-of-art general image completion methods. Our method enables new portrait image editing applications such as occlusion removal and portrait extrapolation. We further show that the proposed general learning framework can be applied to other types of images, e.g. animal images.

cs.GR↗

arXiv2019

On the Role of Geometry in Geo-Localization

Humans can build a mental map of a geographical area to find their way and recognize places. The basic task we consider is geo-localization - finding the pose (position & orientation) of a camera in a large 3D scene from a single image. We aim to experimentally explore the role of geometry in geo-localization in a convolutional neural network (CNN) solution. We do so by ignoring the often available texture of the scene. We therefore deliberately avoid using texture or rich geometric details and use images projected from a simple 3D model of a city, which we term lean images. Lean images contain solely information that relates to the geometry of the area viewed (edges, faces, or relative depth). We find that the network is capable of estimating the camera pose from the lean images, and it does so not by memorization but by some measure of geometric learning of the geographical area. The main contributions of this paper are: (i) providing insight into the role of geometry in the CNN learning process; and (ii) demonstrating the power of CNNs for recovering camera pose using lean images.

cs.CV↗

arXiv2019

GRAINS: Generative Recursive Autoencoders for INdoor Scenes

We present a generative neural network which enables us to generate plausible 3D indoor scenes in large quantities and varieties, easily and highly efficiently. Our key observation is that indoor scene structures are inherently hierarchical. Hence, our network is not convolutional; it is a recursive neural network or RvNN. Using a dataset of annotated scene hierarchies, we train a variational recursive autoencoder, or RvNN-VAE, which performs scene object grouping during its encoding phase and scene generation during decoding. Specifically, a set of encoders are recursively applied to group 3D objects based on support, surround, and co-occurrence relations in a scene, encoding information about object spatial properties, semantics, and their relative positioning with respect to other objects in the hierarchy. By training a variational autoencoder (VAE), the resulting fixed-length codes roughly follow a Gaussian distribution. A novel 3D scene can be generated hierarchically by the decoder from a randomly sampled code from the learned distribution. We coin our method GRAINS, for Generative Recursive Autoencoders for INdoor Scenes. We demonstrate the capability of GRAINS to generate plausible and diverse 3D indoor scenes and compare with existing methods for 3D scene synthesis. We show applications of GRAINS including 3D scene modeling from 2D layouts, scene editing, and semantic scene segmentation via PointNet whose performance is boosted by the large quantity and variety of 3D scenes generated by our method.

cs.GR↗

arXiv2018

Designing Volumetric Truss Structures

We present the first algorithm for designing volumetric Michell Trusses. Our method uses a parametrization approach to generate trusses made of structural elements aligned with the primary direction of an object's stress field. Such trusses exhibit high strength-to-weight ratios. We demonstrate the structural robustness of our designs via a posteriori physical simulation. We believe our algorithm serves as an important complement to existing structural optimization tools and as a novel standalone design tool itself.

cs.GR↗

arXiv2018

Deep Online Video Stabilization

Video stabilization technique is essential for most hand-held captured videos due to high-frequency shakes. Several 2D-, 2.5D- and 3D-based stabilization techniques are well studied, but to our knowledge, no solutions based on deep neural networks had been proposed. The reason for this is mostly the shortage of training data, as well as the challenge of modeling the problem using neural networks. In this paper, we solve the video stabilization problem using a convolutional neural network (ConvNet). Instead of dealing with offline holistic camera path smoothing based on feature matching, we focus on low-latency real-time camera path smoothing without explicitly representing the camera path. Our network, called StabNet, learns a transformation for each input unsteady frame progressively along the time-line, while creating a more stable latent camera path. To train the network, we create a dataset of synchronized steady/unsteady video pairs via a well designed hand-held hardware. Experimental results shows that the proposed online method (without using future frames) performs comparatively to traditional offline video stabilization methods, while running about 30 times faster. Further, the proposed StabNet is able to handle night-time and blurry videos, where existing methods fail in robust feature matching.

cs.GR↗

arXiv2018

Automatic thread painting generation

ThreadTone is an NPR representation of an input image by half-toning using threads on a circle. Current approaches to create ThreadTone paintings greedily draw the chords on the circle. We introduce the concept of chord space, and design a new algorithm to improve the quality of the thread painting. We use an optimization process that estimates the fitness of every chord in the chord space, and an error-diffusion based sampling process that selects a moderate number of chords to produce the output painting. We used an image similarity measure to evaluate the quality of our thread painting and also conducted a user study. Our approach can produce high quality results on portraits, sketches as well as cartoon pictures.

cs.GR↗