How useful is photo-realistic rendering for visual learning?

Data seems cheap to get, and in many ways it is, but the process of creating a high quality labeled dataset from a mass of data is time-consuming and expensive.

cs.CV Computer Science

View graph Source

Paper

References (30)

01What Is Holding Back Convnets for Detection?Bojan Pepik, Rodrigo Benenson, Tobias Ritschel et al.2015 · German Conference on Pattern Recognition · 66 citations In Library

02Ontological supervision for fine grained classification of Street View storefrontsYair Movshovitz-Attias, Qian Yu, Martin C. Stumpe et al.2015 · Computer Vision and Pattern Recognition · 84 citations

03Learning scene-specific pedestrian detectors without real dataHironori Hattori, Vishnu Naresh Boddeti, Kris M. Kitani et al.2015 · Computer Vision and Pattern Recognition · 146 citations

04Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model ViewsHao Su, C. Qi, Yangyan Li et al.2015 · IEEE International Conference on Computer Vision · 752 citations In Library

05Deep Image: Scaling up Image RecognitionRen Wu, Shengen Yan, Yi Shan et al.2015 · arXiv.org · 385 citations In Library

06Going deeper with convolutionsChristian Szegedy, Wei Liu, Yangqing Jia et al.2014 · Computer Vision and Pattern Recognition · 48k citations In Library

07ImageNet Large Scale Visual Recognition ChallengeOlga Russakovsky, Jia Deng, Hao Su et al.2014 · International Journal of Computer Vision · 43k citations In Library

08Virtual and Real World Adaptation for Pedestrian DetectionDavid Vázquez, Antonio M. López, J. Marín et al.2014 · IEEE Transactions on Pattern Analysis and Machine Intelligence · 194 citations

09Beyond PASCAL: A benchmark for 3D object detection in the wildYu Xiang, Roozbeh Mottaghi, S. Savarese2014 · IEEE Winter Conference on Applications of Computer Vision · 845 citations

10Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural NetworksI. Goodfellow, Yaroslav Bulatov, Julian Ibarz et al.2013 · International Conference on Learning Representations · 759 citations In Library

11Correlation Filters for Object AlignmentVishnu Naresh Boddeti, T. Kanade, B. Kumar2013 · 2013 IEEE Conference on Computer Vision and Pattern Recognition · 144 citations

12Large Scale Distributed Deep NetworksJ. Dean, G. Corrado, R. Monga et al.2012 · Neural Information Processing Systems · 4.2k citations

Scroll for more · 18 remaining

Similar papers

Learning for Free: Object Detectors Trained on Synthetic DataCharles Thane MacKay, Teng-Sheng Moh2021 In Library

Physically based synthetic image generation for machine learning: a review of pertinent literatureDominik Schraml2019 · Other Conferences · 23 citations In Library

Metric-Guided Synthetic Image Data Rendering for Deep Learning compatible with Agentic AIMartina Radoynova, Samuel Pantze, Trina De et al.2026 · 0 citations In Library

Improved Sensor Model for Realistic Synthetic Data GenerationKorbinian Hagn, Oliver Grau2021 In Library

Variable photorealistic image synthesis for training dataset generationВадим Санжаров, Vladimir Alexandrovich Frolov, Alexey Gennadievich Voloboy2020 In Library

PUG: Photorealistic and Semantically Controllable Synthetic Data for Representation LearningFlorian Bordes, Shashank Shekhar, Mark Ibrahim et al.2023 · arXiv (Cornell University) · 40 citations In Library

A New Benchmark: On the Utility of Synthetic Data with Blender for Bare Supervised Learning and Downstream Domain AdaptationHui Tang, Kui Jia2023 · arXiv (Cornell University) · 21 citations In Library

Blender as a tool for generating synthetic dataRafał Sieczka, Maciej Pańczyk2020 · Journal of Computer Sciences Institute In Library

PennSyn2Real: Training Object Recognition Models without Human LabelingTy Nguyen, Ian D. Miller, Avraham Cohen et al.2020 · IEEE Robotics and Automation Letters · 9 citations In Library

PennSyn2Real: Training Object Recognition Models Without Human LabelingTy Nguyen, Ian D. Miller, Avi Cohen et al.2021 · IEEE Robotics and Automation Letters In Library

Patents near this paper

METHODS FOR GENERATING A DATASET OF CORRESPONDING IMAGES FOR MACHINE VISION LEARNINGUS 11,120,301 · MASTERCARD INTERNATIONAL INCORPORATED · 2021 GENERATIVE IMAGE SYNTHESIS FOR TRAINING DEEP LEARNING MACHINESUS 10,643,368 · THE BOEING COMPANY · 2020 GENERATIVE IMAGE SYNTHESIS FOR TRAINING DEEP LEARNING MACHINESUS 11,113,864 · THE BOEING COMPANY · 2021 Generation of Synthetic Images For Training a Neural Network ModelUS 10,867,214 · NVIDIA CORPORATION · 2020 VISUAL IMAGE ANNOTATION UTILIZING MACHINE LEARNING FOR IN-TIME FEEDBACKUS 11,049,044 · BLACKSHARK.AI GMBH · 2021

Source of record

arXiv