Research map / Architectures and efficiency / Quantization
Quantization of Fully Convolutional Networks for Accurate Biomedical Image Segmentation
Xiaowei Xu, Qing Lu, Lin Yang, Sharon Hu, Danny Chen, Yu Hu, Yiyu Shi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add INQ-7bit quantization to suggestive annotation and network training | Suggestive annotation pipeline with five FCNs · MICCAI Gland Part A | F1 score: 0.921 -> 0.930 | Improved |
| Add INQ-7bit quantization to suggestive annotation and network training | Suggestive annotation pipeline with five FCNs · MICCAI Gland Part A | Object Dice: 0.904 -> 0.914 | Improved |
| Add INQ-7bit quantization to suggestive annotation and network training | Suggestive annotation pipeline with five FCNs · MICCAI Gland Part B | F1 score: 0.855 -> 0.862 | Improved |
| Add INQ-7bit quantization to suggestive annotation and network training | Suggestive annotation pipeline with five FCNs · MICCAI Gland Part B | Object Dice: 0.858 -> 0.859 | Improved |
| Add INQ-7bit quantization to suggestive annotation and network training | Suggestive annotation pipeline with five FCNs · MICCAI Gland Part A | Object Hausdorff: 44.736 -> 41.783 | Improved |
| Add INQ-7bit quantization to suggestive annotation and network training | Suggestive annotation pipeline with five FCNs · MICCAI Gland Part B | Object Hausdorff: 96.976 -> 97.390 | Hurt |
| Use INQ-7bit instead of floating-point or INQ-5bit quantization (ablation) | Suggestive annotation FCNs · MICCAI Gland | accuracy | Improved |
| Quantize network training rather than train with floating-point representation (ablation) | Segmentation FCNs · MICCAI Gland | accuracy | Hurt |
| Use INQ with 3 bits, DoReFa-Net, or TWN instead of floating point (ablation) | Segmentation FCNs · MICCAI Gland validation | accuracy | Hurt |
| Vary the number of parallel suggestive FCNs (ablation) | Suggestive annotation · MICCAI Gland | accuracy | Mixed |
| Add quantization to suggestive annotation (ablation) | Suggestive annotation FCNs · MICCAI Gland | uncertainty | Improved |
Closest papers by meaning
- Biomedical Image Segmentation via Representative Annotation (AAAI 2019 · similarity 0.89)
- nnWNet: Rethinking the Use of Transformers in Biomedical Image Segmentation and Calling for a Unified Evaluation Benchmark (CVPR 2025 · similarity 0.88)
- Combining Fully Convolutional and Recurrent Neural Networks for 3D Biomedical Image Segmentation (NeurIPS 2016 · similarity 0.88)
- Deep Interactive Segmentation of Medical Images: A Systematic Review and Taxonomy (TPAMI 2024 · similarity 0.87)
- Generative Medical Segmentation (AAAI 2025 · similarity 0.87)
- Difficulty-Aware Attention Network with Confidence Learning for Medical Image Segmentation (AAAI 2019 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
