@@ -11,13 +11,14 @@ The full paper is available at: [https://arxiv.org/abs/1904.01355](https://arxiv
## Highlights
-**Totally anchor-free:** FCOS completely avoids the complicated computation related to anchor boxes and all hyper-parameters of anchor boxes.
-**Memory-efficient:** FCOS uses 2x less training memory footprint than its anchor-based counterpart RetinaNet.
-**Better performance:** The very simple detector achieves better performance (37.1 vs. 36.8) than Faster R-CNN.
-**Better performance:** The very simple one-stage detector achieves much better performance (38.7 vs. 36.8 in AP with ResNet-50) than Faster R-CNN.
-**Faster training:** With the same hardwares, FCOS also requires less training hours (6.5h vs. 8.8h) than Faster R-CNN.
-**State-of-the-art performance:** Without bells and whistles, FCOS achieves state-of-the-art performances.
It achieves **41.5%** (ResNet-101-FPN) and **43.2%** (ResNeXt-64x4d-101) in AP on coco test-dev.
-**State-of-the-art performance:** Our best model based on ResNeXt-64x4d-101 and deformable convolutions achieves **49.0%** in AP on coco test-dev (with multi-scale testing).
## Updates
### 8 August 2019
- FCOS with VoVNet backbones is available at [VoVNet-FCOS](https://github.com/vov-net/VoVNet-FCOS).
### 23 July 2019
- A trick of using a small central region of the BBox for training improves AP by nearly 1 point [as shown here](https://github.com/yqyao/FCOS_PLUS).
...
...
@@ -64,14 +65,15 @@ Once the installation is done, you can follow the below steps to run a quick dem
1) If your model's name is different, please replace `FCOS_R_50_FPN_1x.pth` with your own.
1) If your model's name is different, please replace `FCOS_imprv_R_50_FPN_1x.pth` with your own.
2) If you enounter out-of-memory error, please try to reduce `TEST.IMS_PER_BATCH` to 1.
3) If you want to evaluate a different model, please change `--config-file` to its config file (in [configs/fcos](configs/fcos)) and `MODEL.WEIGHT` to its weights file.
4) Multi-GPU inference is available, please refer to [#78](https://github.com/tianzhi0549/FCOS/issues/78#issuecomment-526990989).
For your convenience, we provide the following trained models (more models are coming soon).
...
...
@@ -79,12 +81,20 @@ For your convenience, we provide the following trained models (more models are c
*All ResNe(x)t based models are trained with 16 images in a mini-batch and frozen batch normalization (i.e., consistent with models in [maskrcnn_benchmark](https://github.com/facebookresearch/maskrcnn-benchmark)).*
Model | Total training mem (GB) | Multi-scale training | Testing time / im | AP (minival) | AP (test-dev) | Link
[1] *1x and 2x mean the model is trained for 90K and 180K iterations, respectively.*\
[2] *We report total training memory footprint on all GPUs instead of the memory footprint per GPU as in maskrcnn-benchmark*. \
[3] *All results are obtained with a single model and without any test time data augmentation such as multi-scale, flipping and etc..*\
[4] *Our results have been improved since our initial release. If you want to check out our original results, please checkout commit [f4fd589](https://github.com/tianzhi0549/FCOS/tree/f4fd58966f45e64608c00b072c801de7f86b4f3a)*. \
[5] *`c128` denotes the model has 128 (instead of 256) channels in towers (i.e., `MODEL.RESNETS.BACKBONE_OUT_CHANNELS` in [config](https://github.com/tianzhi0549/FCOS/blob/master/configs/fcos/fcos_syncbn_bs32_c128_MNV2_FPN_1x.yaml#L10)).*
[2] *All results are obtained with a single model and without any test time data augmentation such as multi-scale, flipping and etc..*\
[3] *`c128` denotes the model has 128 (instead of 256) channels in towers (i.e., `MODEL.RESNETS.BACKBONE_OUT_CHANNELS` in [config](https://github.com/tianzhi0549/FCOS/blob/master/configs/fcos/fcos_syncbn_bs32_c128_MNV2_FPN_1x.yaml#L10)).*\
[4] *The model `FCOS_imprv_dcnv2_X_101_64x4d_FPN_2x` with multi-scale testing achieves 49.0% in AP on COCO test-dev.* Please use `TEST.BBOX_AUG.ENABLED True` to enable multi-scale testing.
## Training
The following command line will train FCOS_R_50_FPN_1x on 8 GPUs with Synchronous Stochastic Gradient Descent (SGD):
...
...
@@ -120,6 +129,7 @@ Note that:
3) If you want to train FCOS with other backbones, please change `--config-file`.
4) The link of ImageNet pre-training X-101-64x4d in the code is invalid. Please download the model [here](https://cloudstor.aarnet.edu.au/plus/s/k3ys35075jmU1RP/download).
5) If you want to train FCOS on your own dataset, please follow this instruction [#54](https://github.com/tianzhi0549/FCOS/issues/54#issuecomment-497558687).
6) Now, training with 8 GPUs and 4 GPUs can have the same performance. Previous performance gap was because we did not synchronize `num_pos` between GPUs when computing loss.
## Contributing to the project
Any pull requests or issues are welcome.
...
...
@@ -135,6 +145,8 @@ Please consider citing our paper in your publications if the project helps your
}
```
# Acknowledgments
We would like to thank [@yqyao](https://github.com/yqyao) for the tricks of center sampling and GIoU. We also thank [@bearcatt](https://github.com/bearcatt) for his suggestion of positioning the center-ness branch with box regression (refer to [#89](https://github.com/tianzhi0549/FCOS/issues/89#issuecomment-516877042)).
