对比卷积与视觉变压器在小规模图像上的效率，发现调优后模型更快更轻。
CNN and ViT Efficiency Study on Tiny ImageNet and DermaMNIST Datasets
- 用ResNet-18做基准，微调四种ViT变体提升效率
- 调优后ViT推理速度更快，参数更少，准确率不降反升
- 适合边缘设备部署，尤其对医疗图像分类有参考价值
本研究评估了卷积神经网络与基于Transformer的架构在医学和通用图像分类任务中的权衡。以ResNet-18为基线，对四种Vision Transformer变体（Tiny、Small、Base、Large）在DermatologyMNIST和TinyImageNet数据集上应用微调策略，目标是在可接受的精度损失下降低推理延迟和模型复杂度。通过系统性超参数调整，我们证明经过适当微调的Vision Transformer可达到或超过基线性能，实现更快推理速度并使用更少参数，凸显其在资源受限环境下的部署可行性。
原文摘要 · Abstract (English)
This study evaluates the trade-offs between convolutional and transformer-based architectures on both medical and general-purpose image classification benchmarks. We use ResNet-18 as our baseline and introduce a fine-tuning strategy applied to four Vision Transformer variants (Tiny, Small, Base, Large) on DermatologyMNIST and TinyImageNet. Our goal is to reduce inference latency and model complexity with acceptable accuracy degradation. Through systematic hyperparameter variations, we demonstrate that appropriately fine-tuned Vision Transformers can match or exceed the baseline's performance, achieve faster inference, and operate with fewer parameters, highlighting their viability for deployment in resource-constrained environments.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。