XTinyU-Net: Training-Free U-Net Scaling via Initialization-Time Sensitivity

• URL: http://arxiv.org/abs/2605.09639v2
• Date: Wed, 13 May 2026 22:20:37 GMT
• Title: XTinyU-Net: Training-Free U-Net Scaling via Initialization-Time Sensitivity
• Abstract summary: We introduce a training-free selection framework to automatically identify dataset-specific U-Net configurations.<n>XTinyU-Net segmentation achieves accuracy comparable to the heavy nnU-Net baseline with 400x-1600x fewer parameters.
• Score: 9.562751934535154
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: While U-Net architectures remain the gold standard for medical image segmentation, their deployment in resource-constrained environments demands aggressive model compression. However, finding an optimally efficient configuration is computationally prohibitive, typically requiring exhaustive train-and-evaluate cycles to find the smallest model that maintains peak performance. In this paper, we introduce a training-free selection framework to automatically identify ultralightweight, dataset-specific U-Net configurations directly at initialization. We observe that systematically scaling down U-Net channel width induces a sharp transition from a stable performance plateau to representational capacity collapse. To pinpoint this boundary without training, we propose a Jacobian-based sensitivity metric that scores discrete, width-capped U-Net variants using a small set of unlabeled images. By analyzing the total variation of this sensitivity curve, we isolate the smallest stable configuration, which we denote as XTinyU-Net. Evaluated across six diverse medical datasets within the nnU-Net framework, XTinyU-Net achieves segmentation accuracy comparable to the heavy nnU-Net baseline with 400x-1600x fewer parameters, and outperforms contemporary lightweight architectures while utilizing 5x-72x fewer parameters. Code is publicly accessible on https://github.com/alvinkimbowa/nntinyunet.git.

Related papers

• Real-Time Semantic Segmentation with Optimized RetinaNet Architectures for Embedded Automotive Systems [0.0]
  This paper presents an optimized semantic segmentation architecture derived from the RetinaNet detection framework.<n> Opt-RetinaSeg replaces the standard ResNet-50 backbone with a hybrid lightweight feature extractor.<n>The proposed model achieves 73.9% mIoU at 70.4 FPS, representing a 7.4x inference speedup and a 4x reduction in model size relative to the ResNet-50 baseline.
  arXiv  Detail & Related papers  (2026-07-21T08:30:00Z)
• Selecting Samples on Graphs: A Unified Dataset Pruning Framework for Lossless Training Acceleration [71.77541419885925]
  We present a unified graph-based DP framework.<n>By modeling the dataset as a weighted graph, DP can be cast as a Maximum Weight Clique Problem (MWCP)<n>Our method outperforms existing DP methods while substantially reducing training cost, reducing training time by over 40% without sacrificing accuracy on ImageNet-1k with ResNet-50.
  arXiv  Detail & Related papers  (2026-06-11T05:13:32Z)
• Dynamic Pre-training: Towards Efficient and Scalable All-in-One Image Restoration [100.54419875604721]
  All-in-one image restoration tackles different types of degradations with a unified model instead of having task-specific, non-generic models for each degradation. We propose DyNet, a dynamic family of networks designed in an encoder-decoder style for all-in-one image restoration tasks. Our DyNet can seamlessly switch between its bulkier and lightweight variants, thereby offering flexibility for efficient model deployment.
  arXiv  Detail & Related papers  (2024-04-02T17:58:49Z)
• GCondNet: A Novel Method for Improving Neural Networks on Small High-Dimensional Tabular Data [14.124731264553889]
  We propose GCondNet to enhance neural networks by leveraging implicit structures present in data. GCondNet exploits the data's high-dimensionality, and thus improves the performance of an underlying predictor network. We demonstrate GCondNet's effectiveness on 12 real-world datasets, where it outperforms 14 standard and state-of-the-art methods.
  arXiv  Detail & Related papers  (2022-11-11T16:13:34Z)
• LilNetX: Lightweight Networks with EXtreme Model Compression and Structured Sparsification [36.651329027209634]
  LilNetX is an end-to-end trainable technique for neural networks. It enables learning models with specified accuracy-rate-computation trade-off.
  arXiv  Detail & Related papers  (2022-04-06T17:59:10Z)
• Dynamic ConvNets on Tiny Devices via Nested Sparsity [3.0313758880048765]
  This work introduces a new training and compression pipeline to build Nested Sparse ConvNets. A Nested Sparse ConvNet consists of a single ConvNet architecture containing N sparse sub-networks with nested weights subsets. Tested on image classification and object detection tasks on an off-the-shelf ARM-M7 Micro Controller Unit.
  arXiv  Detail & Related papers  (2022-03-07T12:07:02Z)
• An Adaptive Device-Edge Co-Inference Framework Based on Soft Actor-Critic [72.35307086274912]
  High-dimension parameter model and large-scale mathematical calculation restrict execution efficiency, especially for Internet of Things (IoT) devices. We propose a new Deep Reinforcement Learning (DRL)-Soft Actor Critic for discrete (SAC-d), which generates the emphexit point, emphexit point, and emphcompressing bits by soft policy iterations. Based on the latency and accuracy aware reward design, such an computation can well adapt to the complex environment like dynamic wireless channel and arbitrary processing, and is capable of supporting the 5G URL
  arXiv  Detail & Related papers  (2022-01-09T09:31:50Z)
• DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers [105.74546828182834]
  We show a hardware-efficient dynamic inference regime, named dynamic weight slicing, which adaptively slice a part of network parameters for inputs with diverse difficulty levels. We present dynamic slimmable network (DS-Net) and dynamic slice-able network (DS-Net++) by input-dependently adjusting filter numbers of CNNs and multiple dimensions in both CNNs and transformers.
  arXiv  Detail & Related papers  (2021-09-21T09:57:21Z)
• Post-training deep neural network pruning via layer-wise calibration [70.65691136625514]
  We propose a data-free extension of the approach for computer vision models based on automatically-generated synthetic fractal images. When using real data, we are able to get a ResNet50 model on ImageNet with 65% sparsity rate in 8-bit precision in a post-training setting.
  arXiv  Detail & Related papers  (2021-04-30T14:20:51Z)
• MSCFNet: A Lightweight Network With Multi-Scale Context Fusion for Real-Time Semantic Segmentation [27.232578592161673]
  We devise a novel lightweight network using a multi-scale context fusion scheme (MSCFNet) The proposed MSCFNet contains only 1.15M parameters, achieves 71.9% Mean IoU and can run at over 50 FPS on a single Titan XP GPU configuration.
  arXiv  Detail & Related papers  (2021-03-24T08:28:26Z)
• Bottleneck Transformers for Visual Recognition [97.16013761605254]
  We present BoTNet, a powerful backbone architecture that incorporates self-attention for vision tasks. We present models that achieve a strong performance of 84.7% top-1 accuracy on the ImageNet benchmark. We hope our simple and effective approach will serve as a strong baseline for future research in self-attention models for vision.
  arXiv  Detail & Related papers  (2021-01-27T18:55:27Z)
• EfficientNet-eLite: Extremely Lightweight and Efficient CNN Models for Edge Devices by Network Candidate Search [13.467017642143583]
  We propose a novel of Network Candidate Search (NCS) to study the trade-off between the resource usage and the performance. In our experiment, we collect candidate CNN models from EfficientNet-B0 to be scaled down in varied way through width, depth, input resolution and compound scaling down. For further embracing the CNN edge application with Application-Specific Integrated Circuit (ASIC), we adjust the architectures of EfficientNet-eLite to build the more hardware-friendly version, EfficientNet-HF.
  arXiv  Detail & Related papers  (2020-09-16T01:11:10Z)

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.