commit

.gitignore

@@ -3,7 +3,6 @@
 /m2_annotations
 evaluation/spice/*
 *.pyc
-*.jar
 /saved_transformer_models
 /tensorboard_logs
 /visualization

README.md

@@ -36,11 +36,10 @@ python -m spacy download en_core_web_md
 
 ## Data Preparation
 
-* **Annotation**. Download the annotation file [annotation.zip](https://drive.google.com/file/d/1Zc2P3-MIBg3JcHT1qKeYuQt9CnQcY5XJ/view?usp=sharing) [1]. Extract and put it in the project root directory.
-* **Feature**. Download processed image features [ResNeXt-101](https://stduestceducn-my.sharepoint.com/:f:/g/personal/zhn_std_uestc_edu_cn/EssZY4Xdb0JErCk0A1Yx3vUBaRbXau88scRvYw4r1ZuwPg?e=f2QFGp) and [ResNeXt-152](https://stduestceducn-my.sharepoint.com/:f:/g/personal/zhn_std_uestc_edu_cn/EssZY4Xdb0JErCk0A1Yx3vUBaRbXau88scRvYw4r1ZuwPg?e=f2QFGp) features [2], put it in the project root directory.
+* **Annotation**. Download the annotation file [m2_annotations](https://drive.google.com/file/d/12EdMHuwLjHZPAMRJNrt3xSE2AMf7Tz8y/view?usp=sharing) [1]. Extract and put it in the project root directory.
+* **Feature**. Download processed image features [ResNeXt-101](https://pan.baidu.com/s/1avz9zaQ7c36XfVFK3ZZZ5w) and [ResNeXt-152](https://pan.baidu.com/s/1avz9zaQ7c36XfVFK3ZZZ5w) features [2] (code `9vtB`), put it in the project root directory.
 <!-- * **Evaluation**. Download the evaluation tools [here](https://pan.baidu.com/s/1xVZO7t8k4H_l3aEyuA-KXQ). Acess code: jcj6. Extarct and put it in the project root directory. -->
 
-
 ## Training
 Run `python train_transformer.py` using the following arguments:
 
@@ -83,7 +82,6 @@ We provide pretrained model [here](https://drive.google.com/file/d/1Y133r4Wd9edi
 |   Reproduced Model (ResNext101) 	|   81.2   	| 39.9 	| 29.6 	|   59.1  	|  133.7 	|  23.3|
 
 
-
 ### Online Evaluation
 We also report the performance of our model on the online COCO test server with an ensemble of four S<sup>2</sup> models. The detailed online test code can be obtained in this [repo](https://github.com/zhangxuying1004/RSTNet).
 
@@ -95,8 +93,8 @@ Huang, and Rongrong Ji. Rstnet: Captioning with adaptive attention on visual and
 ### Citation
 ```
 @inproceedings{S2,
-  author    = {Pengpeng Zeng and
-               Haonan Zhang and
+  author    = {Pengpeng Zeng* and
+               Haonan Zhang* and
                Jingkuan Song and 
                Lianli Gao},
   title     = {S2 Transformer for Image Captioning},
@@ -107,4 +105,4 @@ Huang, and Rongrong Ji. Rstnet: Captioning with adaptive attention on visual and
 ```
 ## Acknowledgements
 Thanks Zhang _et.al_ for releasing the visual features (ResNeXt-101 and ResNeXt-152). Our code implementation is also based on their [repo](https://github.com/zhangxuying1004/RSTNet).   
-Thanks for the original annotations prepared by [M<sup>2</sup> Transformer](https://github.com/aimagelab/meshed-memory-transformer), and effective visual representation from [grid-feats-vqa](https://github.com/facebookresearch/grid-feats-vqa).
+Thanks for the original annotations prepared by [M<sup>2</sup> Transformer](https://github.com/aimagelab/meshed-memory-transformer), and effective visual representation from [grid-feats-vqa](https://github.com/facebookresearch/grid-feats-vqa).

test_transformer.py

@@ -49,12 +49,12 @@ def predict_captions(model, dataloader, text_field):
     device = torch.device('cuda')
 
     parser = argparse.ArgumentParser(description='Transformer')
-    parser.add_argument('--batch_size', type=int, default=10)
-    parser.add_argument('--workers', type=int, default=4)
+    parser.add_argument('--batch_size', type=int, default=50)
+    parser.add_argument('--workers', type=int, default=12)
     parser.add_argument('--m', type=int, default=40)
 
-    parser.add_argument('--features_path', type=str, default='/home/zhanghaonan/RSTNet-master/X101-features/X101_grid_feats_coco_trainval.hdf5')
-    parser.add_argument('--annotation_folder', type=str, default='/home/zhanghaonan/RSTNet-master/m2_annotations')
+    parser.add_argument('--features_path', type=str, default='./X101-features/X101_grid_feats_coco_trainval.hdf5')
+    parser.add_argument('--annotation_folder', type=str, default='./m2_annotations')
 
     # the path of tested model and vocabulary
     parser.add_argument('--model_path', type=str, default='saved_transformer_models/demo_rl_v5_best_test.pth')