By design, average precision (AP) for object detection aims to treat all classes independently: AP is computed independently per category and averaged. On the one hand, this is desirable as it treats all classes, rare to frequent, equally. On the oth
er hand, it ignores cross-category confidence calibration, a key property in real-world use cases. Unfortunately, we find that on imbalanced, large-vocabulary datasets, the default implementation of AP is neither category independent, nor does it directly reward properly calibrated detectors. In fact, we show that the default implementation produces a gameable metric, where a simple, nonsensical re-ranking policy can improve AP by a large margin. To address these limitations, we introduce two complementary metrics. First, we present a simple fix to the default AP implementation, ensuring that it is truly independent across categories as originally intended. We benchmark recent advances in large-vocabulary detection and find that many reported gains do not translate to improvements under our new per-class independent evaluation, suggesting recent improvements may arise from difficult to interpret changes to cross-category rankings. Given the importance of reliably benchmarking cross-category rankings, we consider a pooled version of AP (AP-pool) that rewards properly calibrated detectors by directly comparing cross-category rankings. Finally, we revisit classical approaches for calibration and find that explicitly calibrating detectors improves state-of-the-art on AP-pool by 1.7 points.
Appearance-based detectors achieve remarkable performance on common scenes, but tend to fail for scenarios lack of training data. Geometric motion segmentation algorithms, however, generalize to novel scenes, but have yet to achieve comparable perfor
mance to appearance-based ones, due to noisy motion estimations and degenerate motion configurations. To combine the best of both worlds, we propose a modular network, whose architecture is motivated by a geometric analysis of what independent object motions can be recovered from an egomotion field. It takes two consecutive frames as input and predicts segmentation masks for the background and multiple rigidly moving objects, which are then parameterized by 3D rigid transformations. Our method achieves state-of-the-art performance for rigid motion segmentation on KITTI and Sintel. The inferred rigid motions lead to a significant improvement for depth and scene flow estimation. At the time of submission, our method ranked 1st on KITTI scene flow leaderboard, out-performing the best published method (scene flow error: 4.89% vs 6.31%).
Monocular object detection and tracking have improved drastically in recent years, but rely on a key assumption: that objects are visible to the camera. Many offline tracking approaches reason about occluded objects post-hoc, by linking together trac
klets after the object re-appears, making use of reidentification (ReID). However, online tracking in embodied robotic agents (such as a self-driving vehicle) fundamentally requires object permanence, which is the ability to reason about occluded objects before they re-appear. In this work, we re-purpose tracking benchmarks and propose new metrics for the task of detecting invisible objects, focusing on the illustrative case of people. We demonstrate that current detection and tracking systems perform dramatically worse on this task. We introduce two key innovations to recover much of this performance drop. We treat occluded object detection in temporal sequences as a short-term forecasting challenge, bringing to bear tools from dynamic sequence prediction. Second, we build dynamic models that explicitly reason in 3D, making use of observations produced by state-of-the-art monocular depth estimation networks. To our knowledge, ours is the first work to demonstrate the effectiveness of monocular depth estimation for the task of tracking and detecting occluded objects. Our approach strongly improves by 11.4% over the baseline in ablations and by 5.0% over the state-of-the-art in F1 score.
Embodied perception refers to the ability of an autonomous agent to perceive its environment so that it can (re)act. The responsiveness of the agent is largely governed by latency of its processing pipeline. While past work has studied the algorithmi
c trade-off between latency and accuracy, there has not been a clear metric to compare different methods along the Pareto optimal latency-accuracy curve. We point out a discrepancy between standard offline evaluation and real-time applications: by the time an algorithm finishes processing a particular frame, the surrounding world has changed. To these ends, we present an approach that coherently integrates latency and accuracy into a single metric for real-time online perception, which we refer to as streaming accuracy. The key insight behind this metric is to jointly evaluate the output of the entire perception stack at every time instant, forcing the stack to consider the amount of streaming data that should be ignored while computation is occurring. More broadly, building upon this metric, we introduce a meta-benchmark that systematically converts any single-frame task into a streaming perception task. We focus on the illustrative tasks of object detection and instance segmentation in urban video streams, and contribute a novel dataset with high-quality and temporally-dense annotations. Our proposed solutions and their empirical analysis demonstrate a number of surprising conclusions: (1) there exists an optimal sweet spot that maximizes streaming accuracy along the Pareto optimal latency-accuracy curve, (2) asynchronous tracking and future forecasting naturally emerge as internal representations that enable streaming perception, and (3) dynamic scheduling can be used to overcome temporal aliasing, yielding the paradoxical result that latency is sometimes minimized by sitting idle and doing nothing.
When building a geometric scene understanding system for autonomous vehicles, it is crucial to know when the system might fail. Most contemporary approaches cast the problem as depth regression, whose output is a depth value for each pixel. Such appr
oaches cannot diagnose when failures might occur. One attractive alternative is a deep Bayesian network, which captures uncertainty in both model parameters and ambiguous sensor measurements. However, estimating uncertainties is often slow and the distributions are often limited to be uni-modal. In this paper, we recast the continuous problem of depth regression as discrete binary classification, whose output is an un-normalized distribution over possible depths for each pixel. Such output allows one to reliably and efficiently capture multi-modal depth distributions in ambiguous cases, such as depth discontinuities and reflective surfaces. Results on standard benchmarks show that our method produces accurate depth predictions and significantly better uncertainty estimations than prior art while running near real-time. Finally, by making use of uncertainties of the predicted distribution, we significantly reduce streak-like artifacts and improves accuracy as well as memory efficiency in 3D map reconstruction.
Joint vision and language tasks like visual question answering are fascinating because they explore high-level understanding, but at the same time, can be more prone to language biases. In this paper, we explore the biases in the MovieQA dataset and
propose a strikingly simple model which can exploit them. We find that using the right word embedding is of utmost importance. By using an appropriately trained word embedding, about half the Question-Answers (QAs) can be answered by looking at the questions and answers alone, completely ignoring narrative context from video clips, subtitles, and movie scripts. Compared to the best published papers on the leaderboard, our simple question + answer only model improves accuracy by 5% for video + subtitle category, 5% for subtitle, 15% for DVS and 6% higher for scripts.
Computer vision has undergone a dramatic revolution in performance, driven in large part through deep features trained on large-scale supervised datasets. However, much of these improvements have focused on static image analysis; video understanding
has seen rather modest improvements. Even though new datasets and spatiotemporal models have been proposed, simple frame-by-frame classification methods often still remain competitive. We posit that current video datasets are plagued with implicit biases over scene and object structure that can dwarf variations in temporal structure. In this work, we build a video dataset with fully observable and controllable object and scene bias, and which truly requires spatiotemporal understanding in order to be solved. Our dataset, named CATER, is rendered synthetically using a library of standard 3D objects, and tests the ability to recognize compositions of object movements that require long-term reasoning. In addition to being a challenging dataset, CATER also provides a plethora of diagnostic tools to analyze modern spatiotemporal video architectures by being completely observable and controllable. Using CATER, we provide insights into some of the most recent state of the art deep video architectures.
We address the task of unsupervised retargeting of human actions from one video to another. We consider the challenging setting where only a few frames of the target is available. The core of our approach is a conditional generative model that can tr
anscode input skeletal poses (automatically extracted with an off-the-shelf pose estimator) to output target frames. However, it is challenging to build a universal transcoder because humans can appear wildly different due to clothing and background scene geometry. Instead, we learn to adapt - or personalize - a universal generator to the particular human and background in the target. To do so, we make use of meta-learning to discover effective strategies for on-the-fly personalization. One significant benefit of meta-learning is that the personalized transcoder naturally enforces temporal coherence across its generated frames; all frames contain consistent clothing and background geometry of the target. We experiment on in-the-wild internet videos and images and show our approach improves over widely-used baselines for the task.
Detecting and segmenting individual objects, regardless of their category, is crucial for many applications such as action detection or robotic interaction. While this problem has been well-studied under the classic formulation of spatio-temporal gro
uping, state-of-the-art approaches do not make use of learning-based methods. To bridge this gap, we propose a simple learning-based approach for spatio-temporal grouping. Our approach leverages motion cues from optical flow as a bottom-up signal for separating objects from each other. Motion cues are then combined with appearance cues that provide a generic objectness prior for capturing the full extent of objects. We show that our approach outperforms all prior work on the benchmark FBMS dataset. One potential worry with learning-based methods is that they might overfit to the particular type of objects that they have been trained on. To address this concern, we propose two new benchmarks for generic, moving object detection, and show that our model matches top-down methods on common categories, while significantly out-performing both top-down and bottom-up methods on never-before-seen categories.
While deep feature learning has revolutionized techniques for static-image understanding, the same does not quite hold for video processing. Architectures and optimization techniques used for video are largely based off those for static images, poten
tially underutilizing rich video information. In this work, we rethink both the underlying network architecture and the stochastic learning paradigm for temporal data. To do so, we draw inspiration from classic theory on linear dynamic systems for modeling time series. By extending such models to include nonlinear mappings, we derive a series of novel recurrent neural networks that sequentially make top-down predictions about the future and then correct those predictions with bottom-up observations. Predictive-corrective networks have a number of desirable properties: (1) they can adaptively focus computation on surprising frames where predictions require large corrections, (2) they simplify learning in that only residual-like corrective terms need to be learned over time and (3) they naturally decorrelate an input data stream in a hierarchical fashion, producing a more reliable signal for learning at each layer of a network. We provide an extensive analysis of our lightweight and interpretable framework, and demonstrate that our model is competitive with the two-stream network on three challenging datasets without the need for computationally expensive optical flow.
