diff --git a/README.md b/README.md
index 597fd75..b6877f2 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,108 @@
 
+
+# Dependencies
+   Matplotlib, PyTorch, TorchVision, OpenCV
+   
+      conda install -c conda-forge matplotlib
+      conda install pytorch torchvision cudatoolkit=10.2 -c pytorch
+      conda install -c conda-forge opencv
+
+## Long-term Learning
+
+* Download [coco](http://cocodataset.org) dataset into folder "data-root", so that it looks like:
+
+      data-root
+      ├──coco
+         ├── annotations
+         │   ├── annotations_trainval2017
+         │   └── image_info_test2017
+         └── images
+             ├── test2017
+             ├── train2017
+             └── val2017
+
+
+* Install coco dataset tools (required by PyTorch).
+
+      conda install -c conda-forge pycocotools
+
+* Run
+
+      python3 train_coder.py --data-root [data-root] --model-save saves/ae.pt
+      
+      # This requires a long time for training on single GPU.
+      # Create a folder "saves" manually and a model named "ae.pt" will be saved.
+
+* You may skip this step, if you download the pre-trained [at.pt](link).
+
+
+## Short-term Learning
+
+* Dowload the [SubT](http://theairlab.org/dataset/interestingness) front camera data (SubTF) and put into folder "data-root", so that it looks like:
+
+      data-root
+      ├──SubTF
+         ├── 0817-ugv0-tunnel0
+         ├── 0817-ugv1-tunnel0
+         ├── 0818-ugv0-tunnel1
+         ├── 0818-ugv1-tunnel1
+         ├── 0820-ugv0-tunnel1
+         ├── 0821-ugv0-tunnel0
+         ├── 0821-ugv1-tunnel0
+         ├── ground-truth
+         └── train
+
+* Run
+
+      python3 train_interest.py --data-root [data-root] --model-save saves/ae.pt --dataset SubTF --memory-size 1000 --save-flag n1000
+      
+      # This will read the previous model "ae.pt".
+      # A new model "ae.pt.SubTF.n1000.mse" will be generated.
+ 
+* You may skip this step, if you download the pre-trained [ae.pt.SubTF.n1000.mse](link).
+ 
+ 
+ ## On-line Learning
+ 
+ * Run
+ 
+         python3 test_interest.py --data-root [data-root] --model-save saves/ae.pt.SubTF.n1000.mse --dataset SubTF --test-data 0
+
+         # --test-data The sequence ID in the dataset SubTF, [0-6] is avaiable
+         # This will read the trained model "ae.pt.SubTF.n1000.mse" from short-term learning.
+         
+ * For convenience, you may run
+ 
+         bash test.sh
+ 
+ * This will generate results files that are compatible with the evaluation metric in [SubT](https://github.com/wang-chen/SubT.git)
+
+ * You may skip this step, if you download our generated [results](link).
+
+
+# Evaluation
+
+* We follow the [SubT](https://github.com/wang-chen/SubT.git) tutorial for evaluation, simply run
+
+      python performance.py --data-root [data-root] --save-flag n1000 --category interest-1
+      # mean accuracy: [0.66235087 0.84281507 0.95655934]
+ 
+      python performance.py --data-root [data-root] --save-flag n1000 --category interest-2
+      # mean accuracy: [0.40703316 0.58456123 0.76820896]
+      
+* This will generate performance figures and create data curves for two categories in folder "performance".
+
+
+# Citation
+
+      @article{wang2020visual,
+        author = {Wang, Chen and Wang, Wenshan and Qiu, Yuheng and Hu, Yafei and Scherer, Sebastian},
+        journal = {arXiv preprint arXiv:2005.08829},
+        title = {{Visual Memorability for Robotic Interestingness via Unsupervised Online Learning}},
+        year = {2020}
+      }
+
+---
 You may watch the following video to catch the idea of this work.
 
-   [<img src="https://img.youtube.com/vi/gBBdYdUrIcw/maxresdefault.jpg" width="100%">](https://youtu.be/gBBdYdUrIcw)
+[<img src="https://img.youtube.com/vi/gBBdYdUrIcw/maxresdefault.jpg" width="100%">](https://youtu.be/gBBdYdUrIcw)