Data to 14 Oct 2026

On Pre-Trained Image Features and Synthetic Images for Deep Learning

arXiv 1710.10710 ↗First posted 29 Oct 2017cs.CVcited by 51

Not on the map. Why?

Abstract (arXiv)

Deep Learning methods usually require huge amounts of training data to perform at their full potential, and often require expensive manual labeling. Using synthetic images is therefore very attractive to train object detectors, as the labeling comes for free, and several approaches have been proposed to combine synthetic and real images for training. In this paper, we show that a simple trick is sufficient to train very effectively modern object detectors with synthetic images only: We freeze the layers responsible for feature extraction to generic layers pre-trained on real images, and train only the remaining layers with plain OpenGL rendering. Our experiments with very recent deep architectures for object recognition (Faster-RCNN, R-FCN, Mask-RCNN) and image feature extractors (InceptionResnet and Resnet) show this simple approach performs surprisingly well.

Key facts. Cited by 51 papers in the map (data through 14 Oct 2026). First posted on arXiv 29 Oct 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 7; 2022: 17; 2023: 10; 2024: 12; 2025: 2; 2026: 3.

51 papers.

Sort:

By area:

2023Segment Anything in High Qualitycited by 376
2022ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentationcited by 178
2023BOP Challenge 2022 on Detection, Segmentation and Pose Estimation of Specific Rigid Objectscited by 70
2021Deep Regression on Manifolds: A 3D Rotation Case Studycited by 67
2024BOP Challenge 2023 on Detection, Segmentation and Pose Estimation of Seen and Unseen Rigid Objectscited by 60
2022X-Paste: Revisiting Scalable Copy-Paste for Instance Segmentation using CLIP and StableDiffusioncited by 49
2023MosaicFusion: Diffusion Models as Data Augmenters for Large Vocabulary Instance Segmentationcited by 35
2022DPODv2: Dense Correspondence-Based 6 DoF Pose Estimationcited by 32
2022Object Detection Using Sim2Real Domain Randomization for Robotic Applicationscited by 25
2023Markerless Camera-to-Robot Pose Estimation via Self-supervised Sim-to-Real Transfercited by 24
2022EgoEnv: Human-centric environment representations from egocentric videocited by 22
2022DALL-E for Detection: Language-driven Compositional Image Synthesis for Object Detectioncited by 17

38 more

2022Tree Detection and Diameter Estimation Based on Deep Learningcited by 12
2025Enhancing Vision-Language Compositional Understanding with Multimodal Synthetic Datacited by 12
2022Generative Adversarial Networks for Image Super-Resolution: A Surveycited by 11
2022COPE: End-to-end trainable Constant Runtime Object Pose Estimationcited by 10
2023Beyond Generation: Harnessing Text to Image Models for Object Detection and Segmentationcited by 10
2022Evaluating a Synthetic Image Dataset Generated with Stable Diffusioncited by 9
2022Analysis of Training Object Detection Models with Synthetic Datacited by 9
2022Knock, knock. Who's there? -- Identifying football player jersey numbers with synthetic datacited by 7
2024A SAM-guided Two-stream Lightweight Model for Anomaly Detectioncited by 7
2022Photo-realistic Neural Domain Randomizationcited by 5
2023Object Detector Differences when using Synthetic and Real Training Datacited by 5
2023Federated Learning Over Images: Vertical Decompositions and Pre-Trained Backbones Are Difficult to Beatcited by 4
2022Label-Free Synthetic Pretraining of Object Detectorscited by 3
2023HaDR: Applying Domain Randomization for Generating Synthetic Multimodal Dataset for Hand Instance Segmentation in Cluttered Industrial Environmentscited by 3
2024Harnessing Machine Learning for Discerning AI-Generated Synthetic Imagescited by 3
2024Transfer learning with generative models for object detection on limited datasetscited by 3
2024How Important are Data Augmentations to Close the Domain Gap for Object Detection in Orbit?cited by 3
2021A Pipeline for Vision-Based On-Orbit Proximity Operations Using Deep Learning and Synthetic Imagerycited by 2
2021Learning from scarce information: using synthetic data to classify Roman fine ware potterycited by 2
2021Beyond Classification: Knowledge Distillation using Multi-Object Impressionscited by 2
2023Training on Synthetic Data Beats Real Data in Multimodal Relation Extractioncited by 2
2025Improvise, Adapt, Overcome -- Telescopic Adapters for Efficient Fine-tuning of Vision Language Models in Medical Imagingcited by 2
2021A novel method for object detection using deep learning and CAD modelscited by 1
2021DeePaste -- Inpainting for Pastingcited by 1
2022Real-time Detection of 2D Tool Landmarks with Synthetic Training Datacited by 1
2023HazardNet: Road Debris Detection by Augmentation of Synthetic Modelscited by 1
2024Drone Detection using Deep Neural Networks Trained on Pure Synthetic Datacited by 1
2024Paint Outside the Box: Synthesizing and Selecting Training Data for Visual Groundingcited by 1
2021Learning to Segment Human Body Parts with Synthetically Trained Deep Convolutional Networkscited by 0
2022Autoencoder for Synthetic to Real Generalization: From Simple to More Complex Scenescited by 0
2022FSID: Fully Synthetic Image Denoising via Procedural Scene Generationcited by 0
2024Detection and Pose Estimation of flat, Texture-less Industry Objects on HoloLens using synthetic Trainingcited by 0
2024Genetic Learning for Designing Sim-to-Real Data Augmentationscited by 0
2024A comparative analysis of deep learning models for lung segmentation on X-ray imagescited by 0
2024Mitigating Dimensionality in 2D Rectangle Packing Problem under Reinforcement Learning Schemacited by 0
2024Cartesian Genetic Programming Approach for Designing Convolutional Neural Networkscited by 0
2026SynMulti: Synthetic-to-Real Learning for Multimodal Video Understandingcited by 0
2026The Power of Light: Improving Synthetic-to-Real Domain Adaptation through Physically-Based Indirect Illuminationcited by 0

50 most cited of 51 All 51 (JSON)