# FaPN **Repository Path**: xiaohui-ai/FaPN ## Basic Information - **Project Name**: FaPN - **Description**: https://arxiv.org/pdf/2108.07058.pdf https://github.com/EMI-Group/FaPN https://mp.weixin.qq.com/s/Jq5IDYicgo4FHpNFxxQX2Q - **Primary Language**: Unknown - **License**: Apache-2.0 - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2021-09-13 - **Last Updated**: 2021-09-13 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # FaPN: Feature-aligned Pyramid Network for Dense Image Prediction [[arXiv]](https://arxiv.org/pdf/2108.07058.pdf) [[Project Page]](http://www.shihuahuang.cn/fapn/) ```BibTex @inproceedings{ huang2021fapn, title={{FaPN}: Feature-aligned Pyramid Network for Dense Image Prediction}, author={Shihua Huang and Zhichao Lu and Ran Cheng and Cheng He}, booktitle={International Conference on Computer Vision (ICCV)}, year={2021} } ``` ## Overview FaPN vs. FPN | Before vs. After Alignment :-------------------------:|:-------------------------:

This project provides the official implementation for our ICCV2021 paper "[FaPN: Feature-aligned Pyramid Network for Dense Image Prediction](https://arxiv.org/pdf/2108.07058.pdf)" based on [Detectron2](https://github.com/facebookresearch/detectron2). FaPN is a simple yet effective top-down pyramidal architecture to generate multi-scale features for dense image prediction. Comprised of a feature alignment module (FAM) and a feature selection module (FSM), FaPN addresses the issue of feature alignment in the original [FPN](https://arxiv.org/abs/1612.03144), leading to substaintial improvements on various dense prediction tasks, such as object detection, semantic, instance, panoptic segmentation, etc. ## Installation This project is based on [Detectron2](https://github.com/facebookresearch/detectron2), which can be constructed as follows. * Install Detectron2 following [the instructions](https://detectron2.readthedocs.io/tutorials/install.html). * Setup the dataset following [the structure](https://github.com/facebookresearch/detectron2/blob/master/datasets/README.md). * Copy this project to `/path/to/detectron2` * Install DCNv2 following [Install DCNv2.md](./DCNv2/README.md). ## Training To train a model with 8 GPUs, run: ```bash cd /path/to/detectron2/tools python3 train_net.py --config-file --num-gpus 8 ``` For example, to launch Faster R-CNN training (1x schedule) with ResNet-50 backbone on 8 GPUs, one should execute: ```bash cd /path/to/detectron2/tools python3 train_net.py --config-file ../configs\COCO-Detection\faster_rcnn_R_50_FAN_1x.yaml --num-gpus 8 ``` ## Evaluation To evaluate a pre-trained model with 8 GPUs, run: ```bash cd /path/to/detectron2/tools python3 train_net.py --config-file --num-gpus 8 --eval-only MODEL.WEIGHTS /path/to/model_checkpoint ``` ## Results ### COCO Object Detection #### Faster R-CNN + FaPN:

Name	lr sched	box AP	box APs	box APm	box APl	download
R50	1x	39.2	24.5	43.3	49.1	model \| log
R101	3x	42.8	27.0	46.2	54.9	model \| log

### Cityscapes Semantic Segmentation #### PointRend + FaPN:

Name	lr sched	mask mIoU	mask i_IoU	mask IoU_sup	mask iIoU_sup	download
R50	1x	80.0	61.3	90.6	78.5	model \| log
R101	1x	80.1	62.2	90.8	78.6	model \| log

### COCO Instance Segmentation #### Mask R-CNN + FaPN:

Name	lr sched	mask AP	mask APs	box AP	box APs	download
R50	1x	36.4	18.1	39.8	24.3	model \| log
R101	3x	39.4	20.9	43.8	27.4	model \| log

#### PointRend + FaPN:

Name	lr sched	mask AP	mask APs	box AP	box APs	download
R50	1x	37.6	18.6	39.4	24.2	model \| log

### COCO Panoptic Segmentation #### PanopticFPN + FaPN:

Name	lr sched	PQ	mask mIoU	St PQ	box AP	Th PQ	download
R50	1x	41.1	43.4	32.5	38.7	46.9	model \| log
R101	3x	44.2	45.7	35.0	43.0	53.3	model \| log