Research map / Detection and segmentation / CNN features
Integrally Pre-Trained Transformer Pyramid Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Jointly pre-train the backbone and feature pyramid (ablation) | iTPN · ImageNet-1K | top-1 accuracy: 84.4 -> 85.1 | Helped |
| Jointly pre-train the backbone and feature pyramid (ablation) | iTPN · COCO, Mask R-CNN 1× | box AP: 50.6 -> 51.5 | Helped |
| Jointly pre-train the backbone and feature pyramid (ablation) | iTPN · ADE20K | mIoU: 51.5 -> 51.8 | Helped |
| Load the pre-trained feature pyramid for downstream fine-tuning (ablation) | integrally pre-trained iTPN · COCO, Mask R-CNN 1× | box AP: 51.5 -> 52.1 | Helped |
| Load the pre-trained feature pyramid for downstream fine-tuning (ablation) | integrally pre-trained iTPN · ADE20K | mIoU: 51.8 -> 52.2 | Helped |
| Use C-MLP instead of convolution in the backbone and feature pyramid (ablation) | iTPN without MFM · ImageNet-1K | top-1 accuracy: 84.3 -> 84.9 | Helped |
| Use C-MLP instead of convolution in the backbone and feature pyramid (ablation) | iTPN without MFM · COCO, Mask R-CNN 1× | box AP: 49.8 -> 51.8 | Helped |
| Use C-MLP instead of convolution in the backbone and feature pyramid (ablation) | iTPN without MFM · ADE20K | mIoU: 50.0 -> 51.8 | Helped |
| Add masked feature modeling (MFM) (ablation) | iTPN with convolution instead of C-MLP · ImageNet-1K | top-1 accuracy: 84.3 -> 84.6 | Helped |
| Add masked feature modeling (MFM) (ablation) | iTPN with convolution instead of C-MLP · COCO, Mask R-CNN 1× | box AP: 49.8 -> 50.8 | Helped |
| Add masked feature modeling (MFM) (ablation) | iTPN with convolution instead of C-MLP · ADE20K | mIoU: 50.0 -> 50.7 | Helped |
| Add masked feature modeling (MFM) (ablation) | iTPN with C-MLP · ImageNet-1K | top-1 accuracy: 84.9 -> 85.1 | Helped |
Closest papers by meaning
- Feature Pyramid Networks for Object Detection (CVPR 2017 · similarity 0.88)
- Layer Grafted Pre-training: Bridging Contrastive Learning And Masked Image Modeling For Label-Efficient Representations (ICLR 2023 · similarity 0.87)
- From Prototypes to General Distributions: An Efficient Curriculum for Masked Image Modeling (CVPR 2025 · similarity 0.86)
- Asymmetric Masked Distillation for Pre-Training Small Foundation Models (CVPR 2024 · similarity 0.86)
- Morphing Tokens Draw Strong Masked Image Models (ICLR 2025 · similarity 0.86)
- Pre-training with Random Orthogonal Projection Image Modeling (ICLR 2024 · similarity 0.86)
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.