To realize robots that can understand human instructions and perform meaningful tasks in the near future, it is important to develop learned models that can understand referential language to identify common objects in real-world 3D scenes. In this p
aper, we develop a spatial-language model for a 3D visual grounding problem. Specifically, given a reconstructed 3D scene in the form of a point cloud with 3D bounding boxes of potential object candidates, and a language utterance referring to a target object in the scene, our model identifies the target object from a set of potential candidates. Our spatial-language model uses a transformer-based architecture that combines spatial embedding from bounding-box with a finetuned language embedding from DistilBert and reasons among the objects in the 3D scene to find the target object. We show that our model performs competitively on visio-linguistic datasets proposed by ReferIt3D. We provide additional analysis of performance in spatial reasoning tasks decoupled from perception noise, the effect of view-dependent utterances in terms of accuracy, and view-point annotations for potential robotics applications.
We have observed significant progress in visual navigation for embodied agents. A common assumption in studying visual navigation is that the environments are static; this is a limiting assumption. Intelligent navigation may involve interacting with
the environment beyond just moving forward/backward and turning left/right. Sometimes, the best way to navigate is to push something out of the way. In this paper, we study the problem of interactive navigation where agents learn to change the environment to navigate more efficiently to their goals. To this end, we introduce the Neural Interaction Engine (NIE) to explicitly predict the change in the environment caused by the agents actions. By modeling the changes while planning, we find that agents exhibit significant improvements in their navigational capabilities. More specifically, we consider two downstream tasks in the physics-enabled, visually rich, AI2-THOR environment: (1) reaching a target while the path to the target is blocked (2) moving an object to a target location by pushing it. For both tasks, agents equipped with an NIE significantly outperform agents without the understanding of the effect of the actions indicating the benefits of our approach.
Long-term planning poses a major difficulty to many reinforcement learning algorithms. This problem becomes even more pronounced in dynamic visual environments. In this work we propose Hierarchical Planning and Reinforcement Learning (HIP-RL), a meth
od for merging the benefits and capabilities of Symbolic Planning with the learning abilities of Deep Reinforcement Learning. We apply HIPRL to the complex visual tasks of interactive question answering and visual semantic planning and achieve state-of-the-art results on three challenging datasets all while taking fewer steps at test time and training in fewer iterations. Sample results can be found at youtu.be/0TtWJ_0mPfI
Existing online 3D shape repositories contain thousands of 3D models but lack photorealistic appearance. We present an approach to automatically assign high-quality, realistic appearance models to large scale 3D shape collections. The key idea is to
jointly leverage three types of online data -- shape collections, material collections, and photo collections, using the photos as reference to guide assignment of materials to shapes. By generating a large number of synthetic renderings, we train a convolutional neural network to classify materials in real photos, and employ 3D-2D alignment techniques to transfer materials to different parts of each shape model. Our system produces photorealistic, relightable, 3D shapes (PhotoShapes).
Imagining a scene described in natural language with realistic layout and appearance of entities is the ultimate test of spatial, visual, and semantic world knowledge. Towards this goal, we present the Composition, Retrieval, and Fusion Network (CRAF
T), a model capable of learning this knowledge from video-caption data and applying it while generating videos from novel captions. CRAFT explicitly predicts a temporal-layout of mentioned entities (characters and objects), retrieves spatio-temporal entity segments from a video database and fuses them to generate scene videos. Our contributions include sequential training of components of CRAFT while jointly modeling layout and appearances, and losses that encourage learning compositional representations for retrieval. We evaluate CRAFT on semantic fidelity to caption, composition consistency, and visual quality. CRAFT outperforms direct pixel generation approaches and generalizes well to unseen captions and to unseen video databases with no text annotations. We demonstrate CRAFT on FLINTSTONES, a new richly annotated video-caption dataset with over 25000 videos. For a glimpse of videos generated by CRAFT, see https://youtu.be/688Vv86n0z8.
We present a scalable approach for Detecting Objects by transferring Common-sense Knowledge (DOCK) from source to target categories. In our setting, the training data for the source categories have bounding box annotations, while those for the target
categories only have image-level annotations. Current state-of-the-art approaches focus on image-level visual or semantic similarity to adapt a detector trained on the source categories to the new target categories. In contrast, our key idea is to (i) use similarity not at the image-level, but rather at the region-level, and (ii) leverage richer common-sense (based on attribute, spatial, etc.) to guide the algorithm towards learning the correct detections. We acquire such common-sense cues automatically from readily-available knowledge bases without any extra human effort. On the challenging MS COCO dataset, we find that common-sense knowledge can substantially improve detection performance over existing transfer-learning baselines.
In this paper, we study the problem of answering visual analogy questions. These questions take the form of image A is to image B as image C is to what. Answering these questions entails discovering the mapping from image A to image B and then extend
ing the mapping to image C and searching for the image D such that the relation from A to B holds for C to D. We pose this problem as learning an embedding that encourages pairs of analogous images with similar transformations to be close together using convolutional neural networks with a quadruple Siamese architecture. We introduce a dataset of visual analogy questions in natural images, and show first results of its kind on solving analogy questions on natural images.
This paper proposes direct learning of image classification from user-supplied tags, without filtering. Each tag is supplied by the user who shared the image online. Enormous numbers of these tags are freely available online, and they give insight ab
out the image categories important to users and to image classification. Our approach is complementary to the conventional approach of manual annotation, which is extremely costly. We analyze of the Flickr 100 Million Image dataset, making several useful observations about the statistics of these tags. We introduce a large-scale robust classification algorithm, in order to handle the inherent noise in these tags, and a calibration procedure to better predict objective annotations. We show that freely available, user-supplied tags can obtain similar or superior results to large databases of costly manual annotations.
