An official website of the United States government
We propose ViC-MAE, a model that combines both Masked AutoEncoders (MAE) and contrastive learning. ViC-MAE is trained using a global representation obtained by pooling the local features learned under an MAE reconstruction loss and using this representation under a contrastive objective across images and video frames. We show that visual representations learned under ViC-MAE generalize well to video and image classification tasks. Particularly, ViC-MAE obtains state-of-the-art transfer learning performance from video to images on Imagenet-1k compared to the recently proposed OmniMAE by achieving a top-1 accuracy of 86% (+1.3% absolute improvement) when trained on the same data and 87.1% (+2.4% absolute improvement) when training on extra data. At the same time, ViC-MAE outperforms most other methods on video benchmarks by obtaining 75.9% top-1 accuracy on the challenging Something something-v2 video benchmark. When training on videos and images from diverse datasets, our method maintains a balanced transfer-learning performance between video and image classification benchmarks, coming only as a close second to the best-supervised method.
more »
« less
Why Is the Video Analytics Accuracy Fluctuating, and What Can We Do About It?
It is a common practice to think of a video as a sequence of images (frames), and re-use deep neural network models that are trained only on images for similar analytics tasks on videos. In this paper, we show that this “leap of faith” that deep learning models that work well on images will also work well on videos is actually flawed.We show that even when a video camera is viewing a scene that is not changing in any humanperceptible way, and we control for external factors like video compression and environment (lighting), the accuracy of video analytics application fluctuates noticeably. These fluctuations occur because successive frames produced by the video camera may look similar visually, but are perceived quite differently by the video analytics applications.We observed that the root cause for these fluctuations is the dynamic camera parameter changes that a video camera automatically makes in order to capture and produce a visually pleasing video. The camera inadvertently acts as an “unintentional adversary” because these slight changes in the image pixel values in consecutive frames, as we show, have a noticeably adverse impact on the accuracy of insights from video analytics tasks that re-use image-trained deep learning models. To address this inadvertent adversarial effect from the camera, we explore the use of transfer learning techniques to improve learning in video analytics tasks through the transfer of knowledge from learning on image analytics tasks. Our experiments with a number of different cameras, and a variety of different video analytics tasks, show that the inadvertent adversarial effect from the camera can be noticeably offset by quickly re-training the deep learning models using transfer learning. In particular, we show that our newly trained Yolov5 model reduces fluctuation in object detection across frames, which leads to better tracking of objects (∼40% fewer mistakes in tracking). Our paper also provides new directions and techniques to mitigate the camera’s adversarial effect on deep learning models used for video analytics applications.
more »
« less
- Award ID(s):
- 2211459
- PAR ID:
- 10420213
- Date Published:
- Journal Name:
- Lecture notes in computer science
- ISSN:
- 0302-9743
- Page Range / eLocation ID:
- 430-448
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)Deep learning now offers state-of-the-art accuracy for many prediction tasks. A form of deep learning called deep convolutional neural networks (CNNs) are especially popular on image, video, and time series data. Due to its high computational cost, CNN inference is often a bottleneck in analytics tasks on such data. Thus, a lot of work in the computer architecture, systems, and compilers communities study how to make CNN inference faster. In this work, we show that by elevating the abstraction level and re-imagining CNN inference as queries , we can bring to bear database-style query optimization techniques to improve CNN inference efficiency. We focus on tasks that perform CNN inference repeatedly on inputs that are only slightly different . We identify two popular CNN tasks with this behavior: occlusion-based explanations (OBE) and object recognition in videos (ORV). OBE is a popular method for “explaining” CNN predictions. It outputs a heatmap over the input to show which regions (e.g., image pixels) mattered most for a given prediction. It leads to many re-inference requests on locally modified inputs. ORV uses CNNs to identify and track objects across video frames. It also leads to many re-inference requests. We cast such tasks in a unified manner as a novel instance of the incremental view maintenance problem and create a comprehensive algebraic framework for incremental CNN inference that reduces computational costs. We produce materialized views of features produced inside a CNN and connect them with a novel multi-query optimization scheme for CNN re-inference. Finally, we also devise novel OBE-specific and ORV-specific approximate inference optimizations exploiting their semantics. We prototype our ideas in Python to create a tool called Krypton that supports both CPUs and GPUs. Experiments with real data and CNNs show that Krypton reduces runtimes by up to 5× (respectively, 35×) to produce exact (respectively, high-quality approximate) results without raising resource requirements.more » « less
-
Monitoring and analysis of wildlife are key to conservation planning and conflict management. The widespread use of camera traps coupled with AI-based analysis tools serves as an excellent example of successful and non-invasive use of technology for design, planning, and evaluation of conservation policies. As opposed to the typical use of camera traps that capture still images or short videos, in this project, we propose to analyze longer term videos monitoring a large flock of birds. This project, which is part of the NSF-TIH Indo-US joint R&D partnership, focuses on solving challenges associated with the analysis of long-term videos captured at feeding grounds and nesting sites, among other such locations that host large flocks of migratory birds. We foresee that the objectives of this project would lead to datasets and benchmarking tools as well as novel algorithms that would be instrumental in developing automated video analysis tools that could in turn help understand individual and social behavior of birds. The first of the key outcomes of this research will include the curation of challenging, real-world datasets for benchmarking various image and video analytics algorithms for tasks such as counting, detection, segmentation, and tracking. Our recent efforts towards this outcome is a curated dataset of 812 high-resolution, point-annotated, images (4K - 32MP) of a flock of Demoiselle cranes (Anthropoides virgo) taken from their feeding site at Khichan, Rajasthan, India. The average number of birds in each image is about 207, with a maximum count of 1500. The benchmark experiments show that state-of-the-art vision techniques struggle with tasks such as segmentation, detection, localization, and density estimation for the proposed dataset. Over the execution of this open science research, we will be scaling this dataset for segmentation and tracking in videos, as well as developing novel techniques for video analytics for wildlife monitoring.more » « less
-
Deep neural networks (DNNs) are vulnerable to adversarial examples—maliciously crafted inputs that cause DNNs to make incorrect predictions. Recent work has shown that these attacks generalize to the physical domain, to create perturbations on physical objects that fool image classifiers under a variety of real-world conditions. Such attacks pose a risk to deep learning models used in safety-critical cyber-physical systems. In this work, we extend physical attacks to more challenging object detection models, a broader class of deep learning algorithms widely used to detect and label multiple objects within a scene. Improving upon a previous physical attack on image classifiers, we create perturbed physical objects that are either ignored or mislabeled by object detection models. We implement a Disappearance Attack, in which we cause a Stop sign to “disappear” according to the detector—either by covering the sign with an adversarial Stop sign poster, or by adding adversarial stickers onto the sign. In a video recorded in a controlled lab environment, the state-of-the-art YOLO v2 detector failed to recognize these adversarial Stop signs in over 85% of the video frames. In an outdoor experiment, YOLO was fooled by the poster and sticker attacks in 72.5% and 63.5% of the video frames respectively. We also use Faster R-CNN, a different object detection model, to demonstrate the transferability of our adversarial perturbations. The created poster perturbation is able to fool Faster R-CNN in 85.9% of the video frames in a controlled lab environment, and 40.2% of the video frames in an outdoor environment. Finally, we present preliminary results with a new Creation Attack, wherein innocuous physical stickers fool a model into detecting nonexistent objects.more » « less
-
The analysis and use of egocentric videos for robotic tasks is made challenging by occlusion due to the hand and the visual mismatch between the human hand and a robot end-effector. In this sense, the human hand presents a nuisance. However, often hands also provide a valuable signal, e.g. the hand pose may suggest what kind of object is being held. In this work, we propose to extract a factored representation of the scene that separates the agent (human hand) and the environment. This alleviates both occlusion and mismatch while preserving the signal, thereby easing the design of models for downstream robotics tasks. At the heart of this factorization is our proposed Video Inpainting via Diffusion Model (VIDM) that leverages both a prior on real-world images (through a large-scale pre-trained diffusion model) and the appearance of the object in earlier frames of the video (through attention). Our experiments demonstrate the effectiveness of VIDM at improving inpainting quality on egocentric videos and the power of our factored representation for numerous tasks: object detection, 3D reconstruction of manipulated objects, and learning of reward functions, policies, and affordances from videos.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
