Dataset Information

Visual Turing test for computer vision systems.

ABSTRACT: Today, computer vision systems are tested by their accuracy in detecting and localizing instances of objects. As an alternative, and motivated by the ability of humans to provide far richer descriptions and even tell a story about an image, we construct a "visual Turing test": an operator-assisted device that produces a stochastic sequence of binary questions from a given test image. The query engine proposes a question; the operator either provides the correct answer or rejects the question as ambiguous; the engine proposes the next question ("just-in-time truthing"). The test is then administered to the computer-vision system, one question at a time. After the system's answer is recorded, the system is provided the correct answer and the next question. Parsing is trivial and deterministic; the system being tested requires no natural language processing. The query engine employs statistical constraints, learned from a training set, to produce questions with essentially unpredictable answers-the answer to a question, given the history of questions and their correct answers, is nearly equally likely to be positive or negative. In this sense, the test is only about vision. The system is designed to produce streams of questions that follow natural story lines, from the instantiation of a unique object, through an exploration of its properties, and on to its relationships with other uniquely instantiated objects.

SUBMITTER: Geman D

PROVIDER: S-EPMC4378453 | biostudies-literature | 2015 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Visual Turing test for computer vision systems.

Geman Donald D Geman Stuart S Hallonquist Neil N Younes Laurent L

Proceedings of the National Academy of Sciences of the United States of America 20150309 12

Today, computer vision systems are tested by their accuracy in detecting and localizing instances of objects. As an alternative, and motivated by the ability of humans to provide far richer descriptions and even tell a story about an image, we construct a "visual Turing test": an operator-assisted device that produces a stochastic sequence of binary questions from a given test image. The query engine proposes a question; the operator either provides the correct answer or rejects the question as ...[more]

PMID: 25755262

Similar Datasets

Project description:BackgroundThe use of computers and other Visual Display Terminal (VDT) screens is increasing in Nepal. However, there is a paucity of evidence on the prevalence of Computer Vision Syndrome (CVS) and other occupational health concerns among employees working in front of VDT screens in the Nepalese population.ObjectivesThis study aims to estimate the prevalence of CVS, musculoskeletal and work-related stress among VDT screen users in the office, as well as their understanding and usage of preventive measures.MethodsThe study was a cross-sectional descriptive study among 319 VDT users in office settings in Kathmandu Metropolitan City, Nepal, using a semi-structured self-administered questionnaire. Multivariate logistic regression analysis was conducted to identify the associated factors at 95% CI. P-value <0.05 was considered as statistically significant.ResultsThe prevalence of CVS was 89.4%. More than eight out of ten study participants reported at least one visual and musculoskeletal symptom. Work-related stress, which was moderate-difficult to handle, was present in 36.7% of the study population. The mean±SD computer usage per day was 7.9±1.9 hours. Tired eye (63.3%), feeling of dry eye (57.8%), headache (56.9%) were the common visual symptoms of CVS reported. Total computer use/day > = 8 hours OR 2.6, improper viewing distance OR 3.2, Not using an anti-glare screen OR 2.6, not using eye-drops, and not wearing protective goggles OR 3.1 were significantly associated with the presence of CVS. There was no statistically significant association between visual symptoms of CVS, musculoskeletal symptoms, and stress with gender.ConclusionCVS was substantially related to not employing preventive measures, working longer hours, and having an incorrect viewing distance. With more hours per day spent in front of a VDT screen, work-related stress and musculoskeletal complaints were also found to be important correlates. Similarly, work-related stress was found more among those who had less than five years of job.

Dataset Information

Visual Turing test for computer vision systems.

Publications

Visual Turing test for computer vision systems.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets