Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/cypw/DPNs
Dual Path Networks
https://github.com/cypw/DPNs
Last synced: about 2 months ago
JSON representation
Dual Path Networks
- Host: GitHub
- URL: https://github.com/cypw/DPNs
- Owner: cypw
- Created: 2017-07-02T00:41:01.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2019-03-13T07:12:27.000Z (over 5 years ago)
- Last Synced: 2024-05-10T11:32:22.401Z (4 months ago)
- Language: Python
- Size: 183 KB
- Stars: 538
- Watchers: 33
- Forks: 148
- Open Issues: 16
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-image-classification - official : https://github.com/cypw/DPNs
- Awesome-MXNet - DPN
- awesome-image-classification - official : https://github.com/cypw/DPNs
README
# Dual Path Networks
This repository contains the code and trained models of:
Yunpeng Chen, Jianan Li, Huaxin Xiao, Xiaojie Jin, Shuicheng Yan, Jiashi Feng. "Dual Path Networks" ([NIPS17](https://arxiv.org/abs/1707.01629)).
- DPNs helped us won the **1st place** in Object Localization Task in [ILSVRC 2017](http://image-net.org/challenges/LSVRC/2017/index), with all competition tasks within Top 3. (Team: [NUS-Qihoo_DPNs](http://image-net.org/challenges/LSVRC/2017/results))
## Implementation
DPNs are implemented by [MXNet \@92053bd](https://github.com/cypw/mxnet/tree/92053bd3e71f687b5315b8412a6ac65eb0cc32d5).
### Augmentation
| Method | Settings |
| :------------- | :--------: |
| Random Mirror | True |
| Random Crop | 8% - 100% |
| Aspect Ratio | 3/4 - 4/3 |
| Random HSL | [20,40,50] |
> Note:
> We did not use PCA Lighting and any other advanced augmentation methods.
> Input images are resized by bicubic interpolation.
### Normalization
The augmented input images are substrated by mean RGB = [ 124, 117, 104 ], and then multiplied by 0.0167.
### Mean-Max Pooling
Here, we introduce a new testing technique by using Mean-Max Pooling which can further improve the performance of a well trained CNN in the testing phase without the need of any training/fine-tuining process. This testing technique is designed for the case when the testing images is larger than training crops. The idea is to first convert a trained CNN model into a [convolutional network](http://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Long_Fully_Convolutional_Networks_2015_CVPR_paper.pdf) and then insert the following Mean-Max Pooling layer (a.k.a. [Max-Avg Pooling](https://arxiv.org/abs/1509.08985)), i.e. 0.5 * (global average pooling + global max pooling), just before the final softmax layer.
Based on our observations, Mean-Max Pooling consistently boost the testing accuracy. We adopted this testing strategy in both LSVRC16 and LSVRC17.
## Results
### ImageNet-1k
**Single Model, Single Crop Validation Error:**
Model
Size
GFLOPs
224x224
320x320
320x320
( with mean-max pooling )
Top 1
Top 5
Top 1
Top 5
Top 1
Top 5
DPN-68
49 MB
2.5
23.57
6.93
22.15
5.90
21.51
5.52
DPN-92
145 MB
6.5
20.73
5.37
19.34
4.66
19.04
4.53
DPN-98
236 MB
11.7
20.15
5.15
18.94
4.44
18.72
4.40
DPN-131
304 MB
16.0
19.93
5.12
18.62
4.23
18.55
4.16
### ImageNet-1k (Pretrained on ImageNet-5k)
**Single Model, Single Crop Validation Error:**
Model
Size
GFLOPs
224x224
320x320
320x320
( with mean-max pooling )
Top 1
Top 5
Top 1
Top 5
Top 1
Top 5
DPN-68
49 MB
2.5
22.45
6.09
20.92
5.26
20.62
5.07
DPN-92
145 MB
6.5
19.98
5.06
19.00
4.37
18.79
4.19
DPN-107
333 MB
18.3
19.75
4.94
18.34
4.19
18.15
4.03
>Note: DPN-107 is not well trained.
### ImageNet-5k
**Single Model, Single Crop Validation Accuracy:**
Model
Size
GFLOPs
224x224
320x320
320x320
( with mean-max pooling )
Top 1
Top 5
Top 1
Top 5
Top 1
Top 5
DPN-68
61 MB
2.5
61.27
85.46
61.54
85.99
62.35
86.20
DPN-92
184 MB
6.5
67.31
89.49
66.84
89.38
67.42
89.76
>Note: The higher model complexity comes from the final classifier. Models trained on ImageNet-5k learn much richer feature representation than models trained on ImageNet-1k.
### Efficiency (Training)
The training speed is tested based on [MXNet \@92053bd](https://github.com/cypw/mxnet/tree/92053bd3e71f687b5315b8412a6ac65eb0cc32d5).
**Multiple Nodes (Without specific code optimization):**
Model | CUDA
/cuDNN | #Node | GPU Card
(per node) | Batch Size
(per GPU) | `kvstore` | GPU Mem
(per GPU) | Training Speed*
(per node)
:-------|:------------:|:----:|:---------------------:|:----------------------:|:---------:|:---------:|:-----------:
DPN-68 | 8.0 / 5.1 | 10 | 4 x K80 (Tesla) | 64 |`dist_sync`| 9337 MiB | 284 img/sec
DPN-92 | 8.0 / 5.1 | 10 | 4 x K80 (Tesla) | 32 |`dist_sync`| 8017 MiB | 133 img/sec
DPN-98 | 8.0 / 5.1 | 10 | 4 x K80 (Tesla) | 32 |`dist_sync`| 11128 MiB | 85 img/sec
DPN-131 | 8.0 / 5.1 | 10 | 4 x K80 (Tesla) | 24 |`dist_sync`| 11448 MiB | 60 img/sec
DPN-107 | 8.0 / 5.1 | 10 | 4 x K80 (Tesla) | 24 |`dist_sync`| 12086 MiB | 55 img/sec
> \*This is the actual training speed, which includes `data augmentation`, `forward`, `backward`, `parameter update`, `network communication`, etc.
> MXNet is awesome, we observed a linear speedup as has been shown in [link](https://github.com/dmlc/mxnet/blob/master/example/image-classification/README.md)
## Trained Models
Model | Size | Dataset | MXNet Model
:--------|:------:|:---------:|:-----------------------------------:
DPN-68 | 49 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/5iCuZ8)
DPN-68\* | 49 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/GZetYA)
DPN-68 | 61 MB |ImageNet-5k|[GoogleDrive](https://goo.gl/FEbhPS)
DPN-92 | 145 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/U4ALbg)
DPN-92 | 138 MB |Places365-Standard|[GoogleDrive](https://goo.gl/fRq1YM)
DPN-92\* | 145 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/1sbov7)
DPN-92 | 184 MB |ImageNet-5k|[GoogleDrive](https://goo.gl/H9shRv)
DPN-98 | 236 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/kjVsLG)
DPN-131 | 304 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/VECv1H)
DPN-107\*| 333 MB |ImageNet-1k|[GoogleDrive](https://goo.gl/YtokAb)
>\*Pretrained on ImageNet-5k and then fine-tuned on ImageNet-1k.
## Third-party Implementations
- [Caffe Implementation](https://github.com/soeaver/caffe-model) **with trained models** by [soeaver](https://github.com/soeaver)
- [Chainer Implementation](https://github.com/oyam/chainer-DPNs) by [oyam](https://github.com/oyam)
- [Keras Implementation](https://github.com/titu1994/Keras-DualPathNetworks) by [titu1994](https://github.com/titu1994)
- [MXNet Implementation](https://github.com/miraclewkf/DPN) by [miraclewkf](https://github.com/miraclewkf)
- [PyTorch Implementation](https://github.com/oyam/pytorch-DPNs) by [oyam](https://github.com/oyam)
- [PyTorch Implementation](https://github.com/rwightman/pytorch-dpn-pretrained) **with trained models** by [rwightman](https://github.com/rwightman)
## Other Resources
ImageNet-1k Trainig/Validation List:
- Download link: [GoogleDrive](https://goo.gl/Ne42bM)
ImageNet-1k category name mapping table:
- Download link: [GoogleDrive](https://goo.gl/YTAED5)
ImageNet-5k Raw Images:
- The ImageNet-5k is a subset of ImageNet10K provided by this [paper](http://vision.stanford.edu/pdf/DengBergLiFei-Fei_ECCV2010.pdf).
- Please download the [ImageNet10K](http://www.image-net.org/download-images) and then extract the ImageNet-5k by the list below.
ImageNet-5k Trainig/Validation List:
- It contains about 5k leaf categories from ImageNet10K. There is no category overlapping between our provided ImageNet-5k and the official ImageNet-1k.
- ~~Download link: [GoogleDrive: https://goo.gl/kNZC4j]~~
- Download link: [GoogleDrive](https://goo.gl/XViHf3)
- Mapping Table: [GoogleDrive](https://goo.gl/vWkHYV)
Places365-Standard Validation List & Matlab code for 10 crops testing:
- Download link: [GoogleDrive](https://goo.gl/jQkMpr)
## Citation
If you use DPN in your research, please cite the paper:
```
@article{Chen2017,
title={Dual Path Networks},
author={Yunpeng Chen, Jianan Li, Huaxin Xiao, Xiaojie Jin, Shuicheng Yan, Jiashi Feng},
journal={arXiv preprint arXiv:1707.01629},
year={2017}
}
```