Abstract Achieving physically consistent image editing remains a significant challenge in computer vision. Existing image editing methods typically rely on neural networks, which struggle to accurately handle shadows and refractions. Conversely, physics-based inverse rendering often requires multi-view optimization, limiting its practicality in single-image scenarios. In this paper, we propose Materialist , a neural-initialized physically based rendering pipeline for single-image inverse rendering. Unlike previous hybrid methods that use physics to guide neural generation, our method leverages neural networks to predict initial material properties, which are then rigorously optimized via progressive differentiable rendering. Our approach enables a range of applications, including material editing, object insertion, and relighting, while also introducing an effective method for editing material transparency via ray-traced refraction without requiring full scene geometry. Furthermore, our envmap estimation method also achieves competitive performance, further enhancing the accuracy of image editing task. Experiments demonstrate strong performance across synthetic and real-world datasets, excelling even on challenging out-of-domain images.

The performance and inference capabilities of neural networks rely heavily on the training data they are exposed to. Generally, larger dataset yield more powerful models. This incentive to continuously extend the training sets of models can lead to data exploitation, where data is being used against the owner's wishes to train neural networks. Even if such misuse of data is suspected, it is currently next to impossible to determine its veracity. This research explores the utilization of adversarial noise to manipulate the performance of neural networks, investigating how those findings can be used to infer whether a collection of data is a member of a training set. It proposes a novel approach to generate “deepmarked” images containing adversarial noise that maximizes its detectability as a training set member, while remaining visually indistinguishable from the original data. The findings of this study demonstrate the feasibility of detecting and inferring the membership status of a data collection within a neural network's training set using the proposed technique, in a restricted black-box setting where the model output only contains the single highest likelihood class.

Learning probabilistic models that can estimate the density of a given set of samples, and generate samples from that density, is one of the fundamental challenges in unsupervised machine learning. We introduce a new generative model based on denoising density estimators (DDEs), which are scalar functions parametrized by neural networks, that are efficiently trained to represent kernel density estimators of the data. Leveraging DDEs, our main contribution is a novel technique to obtain generative models by minimizing the Kullback-Leibler (KL)-divergence directly. We prove that our algorithm for obtaining generative models is guaranteed to converge consistently to the correct solution. Our approach does not require specific network architecture as in normalizing flows (NFs), nor use ordinary differential equation (ODE) solvers as in continuous NFs. Experimental results demonstrate substantial improvement in density estimation and competitive performance in generative model training.

Reducing energy consumption is a critical point for neural network models running on edge devices. In this regard, reducing the number of multiply-accumulate (MAC) operations of Deep Neural Networks (DNNs) running on edge hardware accelerators will reduce the energy consumption during inference. Spiking Neural Networks (SNNs) are an example of bio-inspired techniques that can further save energy by using binary activations, and avoid consuming energy when not spiking. The networks can be configured for equivalent accuracy on a task through DNN-to-SNN conversion frameworks but their conversion is based on rate coding therefore the synaptic operations can be high. In this work, we look into different techniques to enforce sparsity on the neural network activation maps and compare the effect of different training regularizers on the efficiency of the optimized DNNs and SNNs.

Abstract Thanks to recent advancements in image processing and deep learning techniques, visual surface inspection in production lines has become an automated process as long as all the defects are visible in a single or a few images. However, it is often necessary to inspect parts under many different illumination conditions to capture all the defects. Training deep networks to perform this task requires large quantities of annotated data, which are rarely available and cumbersome to obtain. To alleviate this problem, we devised an original augmentation approach that, given a small image collection, generates rotated versions of the images while preserving illumination effects, something that random rotations cannot do. We introduce three real multi-illumination datasets, on which we demonstrate the effectiveness of our illumination preserving rotation approach. Training deep neural architectures with our approach delivers a performance increase of up to 51% in terms of AuPRC score over using standard rotations to perform data augmentation.

Despite the substantial progress made in deep learning in recent years, advanced approaches remain computationally intensive. The trade-off between accuracy and computation time and energy limits their use in real-time applications on low power and other resource-constrained systems. In this paper, we tackle this fundamental challenge by introducing a hybrid optical-digital implementation of a convolutional neural network (CNN) based on engineering of the point spread function (PSF) of an optical imaging system. This is done by coding an imaging aperture such that its PSF replicates a large convolution kernel of the first layer of a pre-trained CNN. As the convolution takes place in the optical domain, it has zero cost in terms of energy consumption and has zero latency independent of the kernel size. Experimental results on two datasets demonstrate that our approach yields more than two orders of magnitude reduction in the computational cost while achieving near-state-of-the-art accuracy, or equivalently, better accuracy at the same computational cost.

We present a novel texture synthesis framework, enabling the generation of infinite, high-quality 3D textures given a 2D exemplar image. Inspired by recent advances in natural texture synthesis, we train deep neural models to generate textures by non-linearly combining learned noise frequencies. To achieve a highly realistic output conditioned on an exemplar patch, we propose a novel loss function that combines ideas from both style transfer and generative adversarial networks. In particular, we train the synthesis network to match the Gram matrices of deep features from a discriminator network. In addition, we propose two architectural concepts and an extrapolation strategy that significantly improve generalization performance. In particular, we inject both model input and condition into hidden network layers by learning to scale and bias hidden activations. Quantitative and qualitative evaluations on a diverse set of exemplars motivate our design decisions and show that our system performs superior to previous state of the art. Finally, we conduct a user study that confirms the benefits of our framework.

Deep neural networks for semantic segmentation are most often trained with RGB color images, which encode the radiation visible to the human eyes. In this paper, we study if additional physical scene information, specifically Near-Infrared (NIR) images, improve the performance of neural networks. NIR information can be captured with conventional silicon-based cameras and provide complementary information to visible images regarding object boundaries and materials. In addition, extending the networks' input from a three to a four channel layer is trivial with respect to changes to the architecture and additional parameters. We perform experiments on several state-of-the-art neural networks trained both on RGB alone and on RGB plus NIR and show that the additional image channel consistently improves semantic segmentation accuracy over conventional RGB input even for powerful architectures.

We present a novel system for sketch-based face image editing, enabling users to edit images intuitively by sketching a few strokes on a region of interest. Our interface features tools to express a desired image manipulation by providing both geometry and color constraints as user-drawn strokes. As an alternative to the direct user input, our proposed system naturally supports a copy-paste mode, which allows users to edit a given image region by using parts of another exemplar image without the need of hand-drawn sketching at all. The proposed interface runs in real-time and facilitates an interactive and iterative workflow to quickly express the intended edits. Our system is based on a novel sketch domain and a convolutional neural network trained end-to-end to automatically learn to render image regions corresponding to the input strokes. To achieve high quality and semantically consistent results we train our neural network on two simultaneous tasks, namely image completion and image translation. To the best of our knowledge, we are the first to combine these two tasks in a unified framework for interactive image editing. Our results show that the proposed sketch domain, network architecture, and training procedure generalize well to real user input and enable high quality synthesis results without additional post-processing.

In this paper we introduce a natural image prior that directly represents a Gaussian-smoothed version of the natural image distribution. We include our prior in a formulation of image restoration as a Bayes estimator that also allows us to solve noise-blind image restoration problems. We show that the gradient of our prior corresponds to the mean-shift vector on the natural image distribution. In addition, we learn the mean-shift vector field using denoising autoencoders, and use it in a gradient descent approach to perform Bayes risk minimization. We demonstrate competitive results for noise-blind deblurring, super-resolution, and demosaicing.