Affordance modeling plays an important role in visual understanding. In this paper, we aim to predict affordances of 3D indoor scenes, specifically what human poses are afforded by a given indoor environment, such as sitting on a chair or standing on the floor. In order to predict valid affordances and learn possible 3D human poses in indoor scenes, we need to understand the semantic and geometric structure of a scene as well as its potential interactions with a human. To learn such a model, a large-scale dataset of 3D indoor affordances is required.

We present a new image dataset generated by the NVIDIA Deep Learning Data Synthesizer intended for use in object detection, pose estimation, and tracking applications. This dataset contains 144k stereo image pairs generated from 18 camera view points of three photorealistic virtual environments with up to 10 objects (chosen randomly from the 21 object models of the YCB dataset) and flying distractors. Object and camera pose, scene lighting, and quantity of objects and distractors were randomized. Each provided view includes RGB, depth, segmentation, and surface normal images.

We introduce an inference technique to produce discriminative context-aware image captions (captions that describe differences between images or visual concepts) using only generic context-agnostic training data (captions that describe a concept or an image in isolation). For example, given images and captions of "siamese cat" and "tiger cat", we generate language that describes the "siamese cat" in a way that distinguishes it from "tiger cat".

We present an approach to effectively use millions of images with noisy annotations in conjunction with a small subset of cleanly-annotated images to learn powerful image representations. One common approach to combine clean and noisy data is to first pre-train a network using the large noisy dataset and then fine-tune with the clean dataset. We show this approach does not fully leverage the information contained in the clean set.

Sharing information among multiple learning agents can accelerate learning. It could be particularly useful if learners operate in continuously changing environments, because a learner could benefit from previous experience of another learner to adapt to their new environment. Such group-adaptive learning has numerous applications, from predicting financial time-series, through content recommendation systems, to visual understanding for adaptive autonomous agents. Here we address the problem in the context of online adaptive learning.

Precomputing high-quality sample points has been shown to be a useful technique for Monte Carlo integration in rendering; doing so allows optimizing properties of the points without the performance constraints of generating samples during rendering. A particularly useful property to incorporate is stratification across elementary intervals, which has been shown to reduce error in Monte Carlo integration. This is a key property of the recently-introduced progressive multi-jittered, pmj02 and pmj02bn points [Christensen et al.

Given a random pair of images, a universal style transfer method extracts the feel from a reference image to synthesize an output based on the look of a content image. Recent algorithms based on second-order statistics, however, are either computationally expensive or prone to generate artifacts due to the trade-off between image quality and run-time performance. In this work, we present an approach for universal style transfer that learns the transformation matrix in a data-driven fashion.

Throughput-oriented architectures, such as GPUs, can sustain three orders of magnitude more concurrent threads than multicore architectures. This level of concurrency pushes typical synchronization primitives (e.g., mutexes) over their scalability limits, creating significant performance bottlenecks in modules, such as memory allocators, that use them.

We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis.