Does YOLO Really Need to See Every Training Image in Every Epoch?

• URL: http://arxiv.org/abs/2603.17684v1
• Date: Wed, 18 Mar 2026 12:59:39 GMT
• Title: Does YOLO Really Need to See Every Training Image in Every Epoch?
• Abstract summary: YOLO detectors are known for their fast inference speed, yet training them remains unexpectedly time-consuming.<n>We propose an Anti-Forgetting Sampling Strategy (AFSS) that dynamically determines which images should be used.<n>AFSS achieves more than $1.43times$ training speedup for YOLO-series detectors while also improving accuracy.
• Score: 58.97843527861311
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: YOLO detectors are known for their fast inference speed, yet training them remains unexpectedly time-consuming due to their exhaustive pipeline that processes every training image in every epoch, even when many images have already been sufficiently learned. This stands in clear contrast to the efficiency suggested by the ``You Only Look Once'' philosophy. This naturally raises an important question: \textit{Does YOLO really need to see every training image in every epoch?} To explore this, we propose an Anti-Forgetting Sampling Strategy (AFSS) that dynamically determines which images should be used and which can be skipped during each epoch, allowing the detector to learn more effectively and efficiently. Specifically, AFSS measures the learning sufficiency of each training image as the minimum of its detection recall and precision, and dynamically categorizes training images into easy, medium, or hard levels accordingly. Easy training images are sparsely resampled during training in a continuous review manner, with priority given to those that have not been used for a long time to reduce redundancy and prevent forgetting. Moderate training images are partially selected, prioritizing recently unused ones and randomly choosing the rest from unselected images to ensure coverage and prevent forgetting. Hard training images are fully sampled in every epoch to ensure sufficient learning. The learning sufficiency of each training image is periodically updated, enabling detectors to adaptively shift its focus toward the informative training images over time while progressively discarding redundant ones. On widely used natural image detection benchmarks (MS COCO 2017 and PASCAL VOC 2007) and remote sensing detection datasets (DOTA-v1.0 and DIOR-R), AFSS achieves more than $1.43\times$ training speedup for YOLO-series detectors while also improving accuracy.

Related papers

• Learning Continuous Source Responses For Generalizable AI-Generated Image Detection [42.338483531657666]
  We introduce CuRe, a framework for learning Continuous Source Responses.<n>CuRe revisits authenticity detection from the perspective of the training task.<n>It achieves an average balanced accuracy of 89.7%, exceeding the second-best method by 5.2 percentage points.
  arXiv  Detail & Related papers  (2026-09-13T06:24:59Z)
• Intermediate Representations are Strong AI-Generated Image Detectors [51.7378156094049]
  We propose a search-based method employing data sensitivity embedding in intermediate layers to detect AI-generated images.<n>Given a set of real and AI-generated images, our method examines the similarity between original image embeddings and perturbed image embeddings, and detects AI-generated images based on the similarity.<n>Our method achieves the largest performance gain on the Forensics Small benchmark by 39.61% compared to the best training-free method and 5.14% compared to the best training-based method in AUROC score.
  arXiv  Detail & Related papers  (2026-05-05T23:26:02Z)
• TTSnap: Test-Time Scaling of Diffusion Models via Noise-Aware Pruning [53.52543819839442]
  A prominent approach to test-time scaling for text-to-image diffusion models formulates the problem as a search over multiple noise seeds.<n>We propose test-time scaling with noise-aware pruning (TTSnap), a framework that prunes low-quality candidates without fully denoising them.
  arXiv  Detail & Related papers  (2025-11-27T09:14:26Z)
• Should VLMs be Pre-trained with Image Data? [54.50406730361859]
  We find that pre-training with a mixture of image and text data allows models to perform better on vision-language tasks.<n>On an average of 6 diverse tasks, we find that for a 1B model, introducing visual tokens 80% of the way through pre-training results in a 2% average improvement over introducing visual tokens to a fully pre-trained model.
  arXiv  Detail & Related papers  (2025-03-10T17:58:19Z)
• EfficientTrain: Exploring Generalized Curriculum Learning for Training Visual Backbones [80.662250618795]
  This paper presents a new curriculum learning approach for the efficient training of visual backbones (e.g., vision Transformers) As an off-the-shelf method, it reduces the wall-time training cost of a wide variety of popular models by >1.5x on ImageNet-1K/22K without sacrificing accuracy.
  arXiv  Detail & Related papers  (2022-11-17T17:38:55Z)
• Enhanced Performance of Pre-Trained Networks by Matched Augmentation Distributions [10.74023489125222]
  We propose a simple solution to address the train-test distributional shift. We combine results for multiple random crops for a test image. This not only matches the train time augmentation but also provides the full coverage of the input image.
  arXiv  Detail & Related papers  (2022-01-19T22:33:00Z)
• OSSID: Online Self-Supervised Instance Detection by (and for) Pose Estimation [17.78557307620686]
  Real-time object pose estimation is necessary for many robot manipulation algorithms. We propose the OSSID framework, leveraging a slow zero-shot pose estimator to self-supervise the training of a fast detection algorithm. We show that this self-supervised training exceeds the performance of existing zero-shot detection methods.
  arXiv  Detail & Related papers  (2022-01-18T20:55:56Z)
• Pure Noise to the Rescue of Insufficient Data: Improving Imbalanced Classification by Training on Random Noise Images [12.91269560135337]
  We present a surprisingly simple yet highly effective method to mitigate this limitation. Unlike the common use of additive noise or adversarial noise for data augmentation, we propose directly training on pure random noise images. We present a new Distribution-Aware Routing Batch Normalization layer (DAR-BN), which enables training on pure noise images in addition to natural images within the same network.
  arXiv  Detail & Related papers  (2021-12-16T11:51:35Z)
• ZSD-YOLO: Zero-Shot YOLO Detection using Vision-Language KnowledgeDistillation [5.424015823818208]
  A dataset such as COCO is extensively annotated across many images but with a sparse number of categories and annotating all object classes across a diverse domain is expensive and challenging. We develop a Vision-Language distillation method that aligns both image and text embeddings from a zero-shot pre-trained model such as CLIP to a modified semantic prediction head from a one-stage detector like YOLOv5. During inference, our model can be adapted to detect any number of object classes without additional training.
  arXiv  Detail & Related papers  (2021-09-24T16:46:36Z)
• Memory Efficient Meta-Learning with Large Images [62.70515410249566]
  Meta learning approaches to few-shot classification are computationally efficient at test time requiring just a few optimization steps or single forward pass to learn a new task. This limitation arises because a task's entire support set, which can contain up to 1000 images, must be processed before an optimization step can be taken. We propose LITE, a general and memory efficient episodic training scheme that enables meta-training on large tasks composed of large images on a single GPU.
  arXiv  Detail & Related papers  (2021-07-02T14:37:13Z)
• Active Ensemble Deep Learning for Polarimetric Synthetic Aperture Radar Image Classification [10.80252725670625]
  In this letter, we take the advantage of active learning and propose active ensemble deep learning (AEDL) for PolSAR image classification. We show that only 35% of the predicted labels of a deep learning model's snapshots near its convergence were exactly the same. Using the snapshots committee to give out the informativeness of unlabeled data, the proposed AEDL achieved better performance on two real PolSAR images compared with standard active learning strategies.
  arXiv  Detail & Related papers  (2020-06-29T01:40:54Z)
• BP-DIP: A Backprojection based Deep Image Prior [49.375539602228415]
  We propose two image restoration approaches: (i) Deep Image Prior (DIP), which trains a convolutional neural network (CNN) from scratch in test time using the degraded image; and (ii) a backprojection (BP) fidelity term, which is an alternative to the standard least squares loss that is usually used in previous DIP works. We demonstrate the performance of the proposed method, termed BP-DIP, on the deblurring task and show its advantages over the plain DIP, with both higher PSNR values and better inference run-time.
  arXiv  Detail & Related papers  (2020-03-11T17:09:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.