Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/HCIILAB/DeRPN

A novel region proposal network for more general object detection ( including scene text detection ).
https://github.com/HCIILAB/DeRPN

detection object-detection object-proposals region-proposals rpn scene-text scene-text-detection text-detection

Last synced: 9 days ago
JSON representation

A novel region proposal network for more general object detection ( including scene text detection ).

Host: GitHub
URL: https://github.com/HCIILAB/DeRPN
Owner: HCIILAB
Created: 2018-11-10T11:57:09.000Z (about 6 years ago)
Default Branch: master
Last Pushed: 2019-03-13T05:24:53.000Z (over 5 years ago)
Last Synced: 2024-08-02T11:15:02.233Z (3 months ago)
Topics: detection, object-detection, object-proposals, region-proposals, rpn, scene-text, scene-text-detection, text-detection
Language: Python
Homepage:
Size: 9.48 MB
Stars: 155
Watchers: 23
Forks: 46
Open Issues: 2
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

        # DeRPN:  Taking a further step toward more general object detection

DeRPN is a novel region proposal network which concentrates on improving the adaptivity of current detectors. 

The paper is available [here](https://arxiv.org/abs/1811.06700).



## Recent Update

**·** Mar. 13, 2019: The DeRPN pretrained models are added.

**·** Jan. 25, 2019: The code is released.

## Contact Us

Welcome to improve DeRPN together. For any questions, please feel free to contact Lele Xie ([email protected]) or Prof. Jin ([email protected]).

## Citation

If you find DeRPN useful to your research, please consider citing our paper as follow:

```

@article{xie2019DeRPN,

  title     = {DeRPN: Taking a further step toward more general object detection},

  author    = {Lele Xie, Yuliang Liu, Lianwen Jin*, Zecheng Xie}

  joural    = {AAAI}

  year      = {2019}

}

```

## Main Results

**Note**: The reimplemented results are slightly different from those presented  in the paper for different training settings, but the conclusions are still consistent. For example, this code doesn't use multi-scale training which should boost the results for both DeRPN and RPN.

### COCO-Text

training data:  COCO-Text train

test data: COCO-Text test

|                   | network | [email protected]  | [email protected] | [email protected] | [email protected] |

|:-----------------:|:---------------------:|:---------------------:|:---------------------:|:---------------------:|:---------------------:|

| RPN+Faster R-CNN | VGG16 | 32.48 | 52.54 | 7.40 | 17.59 |

| DeRPN+Faster R-CNN | VGG16 | 47.39 | 70.46 | 11.05 | 25.12 |

|RPN+R-FCN  |ResNet-101  | 37.71 | 54.35 | 13.17 | 22.21 |

|DeRPN+R-FCN |ResNet-101 | 48.62 | 71.30 | 13.37 | 27.57 |

### Pascal VOC

training data:  VOC 07+12 trainval 

test data: VOC 07 test

Inference time is evaluated on one TITAN XP GPU.

|                     |  network   | inference time | [email protected] | [email protected] |  AP   |

| :-----------------: | :--------: | :------------: | :----: | :-----: | :---: |

|  RPN+Faster R-CNN   |   VGG16    |     64 ms      | 75.53  |  42.08  | 42.60 |

| DeRPN+Faster R-CNN  |   VGG16    |     65 ms      | 76.17  |  44.97  | 43.84 |

|      RPN+R-FCN      | ResNet-101 |     85 ms      | 78.87  |  54.30  | 50.04 |

| DeRPN+R-FCN (900) * | ResNet-101 |     84 ms      | 79.21  |  54.43  | 50.28 |

( "*": On Pascal VOC dataset, we found that it is more suitable to train the DeRPN+R-FCN model with 900 proposals. For other experiments, we use the default proposal number to train the models, i.e., 2000 proposals fro Faster R-CNN, 300 proposals for R-FCN. )

### MS COCO

training data:  COCO 2017 train

test data: COCO 2017 test/val

| test set | network |  AP  | AP50 | AP75 | AP_S  | AP_M  | AP_L  |

| :----------------: | :-----: | :--: | :----: | :-----: | ---- | ---- | :--: |

|  RPN+Faster R-CNN  |  VGG16  | 24.2 | 45.4 | 23.7 | 7.6 | 26.6 | 37.3 |

| DeRPN+Faster R-CNN |  VGG16  | 25.5 | 47.2 | 25.2 | 10.3 | 27.9 | 36.7 |

|     RPN+R-FCN      | ResNet-101 | 27.7 | 47.9 | 29.0 | 10.1 | 30.2 | 40.1 |

|    DeRPN+R-FCN     | ResNet-101 | 28.4 | 49.0 | 29.5 | 11.1 | 31.7 | 40.5 |

| val set | network |  AP  | AP50 | AP75 |AP_S  | AP_M  | AP_L  |

| :----------------: | :-----: | :--: | :----: | :-----: | ---- | ---- | :--: |

|  RPN+Faster R-CNN  |  VGG16  | 24.1 | 45.0 | 23.8 | 7.6            | 27.8           |      37.8      |

| DeRPN+Faster R-CNN |  VGG16  | 25.5 | 47.3 | 25.0 | 9.9 | 28.8 | 37.8 |

|     RPN+R-FCN      | ResNet-101 | 27.8 | 48.1 | 28.8 | 10.4 | 31.2 | 42.5 |

|    DeRPN+R-FCN     | ResNet-101 | 28.4 | 48.5 | 29.5 | 11.5 | 32.9 | 42.0 |

## Getting Started

1. Requirements

2. Installation

3. Preparation for Training & Testing

4. Usage

## Requirements

1. Cuda 8.0 and cudnn 5.1. 

2. Some python packages: cython, opencv-python, easydict et. al. Simply install them if your system misses these packages.

3. Configure the caffe according to your environment ([Caffe installation instructions](http://caffe.berkeleyvision.org/installation.html)). As the code requires pycaffe, caffe should be built with python layers. In Makefile.config, make sure to uncomment this line:

  ```

  WITH_PYTHON_LAYER := 1

  ```

4. An NVIDIA GPU with more than 6GB is required for ResNet-101.

## Installation

1. Clone the DeRPN repository

    ```

    git clone https://github.com/HCIILAB/DeRPN.git

    ```

2. Build the Cython modules

    ```Shell

    cd $DeRPN_ROOT/lib

    make

    ```

3. Build caffe and pycaffe

    ```Shell

    cd $DeRPN_ROOT/caffe

    make -j8 && make pycaffe

    ```

## Preparation for Training & Testing

### Dataset

1. Download the datasets of [Pascal VOC 2007 & 2012](http://host.robots.ox.ac.uk/pascal/VOC/), [MS COCO 2017](http://cocodataset.org/#download) and [COCO-Text](http://rrc.cvc.uab.es/?ch=5&com=introduction).

2. You need to put these datasets under the $DeRPN_ROOT/data folder (with symlinks). 

3. For COCO-Text, the folder structure is as follow:

    ```Shell

    $DeRPN_ROOT/data/coco_text/images/train2014

    $DeRPN_ROOT/data/coco_text/images/val2014

    $DeRPN_ROOT/data/coco_text/annotations  

    # train2014, val2014, and annotations are symlinks from /pth_to_coco2014/train2014, 

    # /pth_to_coco2014/val2014 and /pth_to_coco2014/annotations2014/, respectively.

    ```

4. For COCO, the folder structure is as follow:

    ```Shell

    $DeRPN_ROOT/data/coco/images/train2017

    $DeRPN_ROOT/data/coco/images/val2017

    $DeRPN_ROOT/data/coco/images/test-dev2017

    $DeRPN_ROOT/data/coco/annotations  

    # the symlinks are similar to COCO-Text

    ```

5. For Pascal VOC, the folder structure is as follow:

    ```Shell

    $DeRPN_ROOT/data/VOCdevkit2007

    $DeRPN_ROOT/data/VOCdevkit2012

    #VOCdevkit2007 and VOCdevkit2012 are symlinks from $VOCdevkit whcich contains VOC2007 and VOC2012.

    ```

### Pretrained models

Please download the ImageNet pretrained models ([VGG16](https://pan.baidu.com/s/1BDDl5xtrBznlyIrVj9g_zQ) and [ResNet-101](https://pan.baidu.com/s/1BDDl5xtrBznlyIrVj9g_zQ), password: k4z1), and put them under

```Shell

$DeRPN_ROOT/data/imagenet_models

```

We also provide the DeRPN pretrained models [here](https://pan.baidu.com/s/141Dy0OiXLMau-XLdEkHdIA) (password: fsd8).

## Usage

```Shell

cd $DeRPN_ROOT

./experiments/scripts/faster_rcnn_derpn_end2end.sh [GPU_ID] [NET] [DATASET]

# e.g., ./experiments/scripts/faster_rcnn_derpn_end2end.sh 0 VGG16 coco_text

```

## Copyright

This code is free to the academic community for research purpose only. For commercial purpose usage, please contact Dr. Lianwen Jin: [email protected].