mlsanityData to 14 Oct 2026

Label-Free Synthetic Pretraining of Object Detectors

arXiv 2208.04268 ↗First posted 8 Aug 2022cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

We propose a new approach, Synthetic Optimized Layout with Instance Detection (SOLID), to pretrain object detectors with synthetic images. Our "SOLID" approach consists of two main components: (1) generating synthetic images using a collection of unlabelled 3D models with optimized scene arrangement; (2) pretraining an object detector on "instance detection" task - given a query image depicting an object, detecting all instances of the exact same object in a target image. Our approach does not need any semantic labels for pretraining and allows the use of arbitrary, diverse 3D models. Experiments on COCO show that with optimized data generation and a proper pretraining task, synthetic data can be highly effective data for pretraining object detectors. In particular, pretraining on rendered images achieves performance competitive with pretraining on real images while using significantly less computing resources. Code is available at https://github.com/princeton-vl/SOLID.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 55 of its 77 references that are in the map. First posted on arXiv 8 Aug 2022.

Builds on

55 of 77 references.

References · 55

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  4. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  5. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,513
  6. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  7. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  8. 2016Feature Pyramid Networks for Object Detectioncited by 5,036
  9. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,750
  10. 2015ShapeNet: An Information-Rich 3D Model Repositorycited by 2,897
  11. 2020Exploring Simple Siamese Representation Learningcited by 2,866
  12. 2020Unsupervised Learning of Visual Features by Contrasting Cluster Assignmentscited by 2,857
38 more
  1. 2021BEiT: BERT Pre-Training of Image Transformerscited by 2,375
  2. 2020Improved Baselines with Momentum Contrastive Learningcited by 2,287
  3. 2013Rich feature hierarchies for accurate object detection and semantic segmentationcited by 2,162
  4. 2021An Empirical Study of Training Self-Supervised Vision Transformerscited by 1,616
  5. 2016Context Encoders: Feature Learning by Inpaintingcited by 1,428
  6. 2020Big Self-Supervised Models are Strong Semi-Supervised Learnerscited by 1,380
  7. 2018Deep Clustering for Unsupervised Learning of Visual Featurescited by 1,028
  8. 2019EfficientDet: Scalable and Efficient Object Detectioncited by 1,020
  9. 2015A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimationcited by 881
  10. 2016Playing for Data: Ground Truth from Computer Gamescited by 818
  11. 2019Self-Supervised Learning of Pretext-Invariant Representationscited by 808
  12. 2021DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Camerascited by 687
  13. 2020Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentationcited by 596
  14. 2019Self-labelling via simultaneous clustering and representation learningcited by 528
  15. 2020TartanAir: A Dataset to Push the Limits of Visual SLAMcited by 506
  16. 2020Sparse R-CNN: End-to-End Object Detection with Learnable Proposalscited by 495
  17. 2018Rethinking ImageNet Pre-trainingcited by 377
  18. 2021RAFT-Stereo: Multilevel Recurrent Field Transforms for Stereo Matchingcited by 353
  19. 2018Training Deep Networks with Synthetic Data: Bridging the Reality Gap by Domain Randomizationcited by 294
  20. 2020CosyPose: Consistent multi-view multi-object 6D pose estimationcited by 287
  21. 2020Frustratingly Simple Few-Shot Object Detectioncited by 282
  22. 2019Revisiting Self-Supervised Visual Representation Learningcited by 251
  23. 2019GA-Net: Guided Aggregation Net for End-to-end Stereo Matchingcited by 244
  24. 2019Unsupervised Embedding Learning via Invariant and Spreading Instance Featurecited by 232
  25. 2021DetCo: Unsupervised Contrastive Learning for Object Detectioncited by 205
  26. 2017Multi-task Self-Supervised Visual Learningcited by 200
  27. 2016Driving in the Matrix: Can Virtual Worlds Replace Human-Generated Annotations for Real World Tasks?cited by 194
  28. 2021Efficient Visual Pretraining with Contrastive Detectioncited by 150
  29. 2021Probabilistic two-stage detectioncited by 130
  30. 2019Domain-invariant Stereo Matching Networkscited by 109
  31. 2021Instance Localization for Self-supervised Detection Pretrainingcited by 99
  32. 2021Region Similarity Representation Learningcited by 96
  33. 2018Falling Things: A Synthetic Dataset for 3D Object Detection and Pose Estimationcited by 92
  34. 2019Unsupervised Deep Learning by Neighbourhood Discoverycited by 78
  35. 2021Exponential Moving Average Normalization for Self-supervised and Semi-supervised Learningcited by 78
  36. 2019BlenderProccited by 77
  37. 2021DETReg: Unsupervised Pretraining with Region Priors for Object Detectioncited by 76
  38. 2021Spatially Consistent Representation Learningcited by 64

50 most cited of 55 All 55 (JSON)

Built on by

Cited by year

Citing papers per year: 2023: 2; 2024: 1.

3 papers.

  1. 2023Infinite Photorealistic Worlds using Procedural Generationcited by 147
  2. 2024Infinigen Indoors: Photorealistic Indoor Scenes using Procedural Generationcited by 137
  3. 2023View-Dependent Octree-based Mesh Extraction in Unbounded Scenes for Procedural Synthetic Datacited by 0