Update README with usage instructions
Browse files- README.md +49 -0
- assets/images/datasets-tab.JPG +0 -0
- assets/images/find-expressions-tab.JPG +0 -0
README.md
CHANGED
@@ -13,3 +13,52 @@ license: mit
|
|
13 |
|
14 |
# IndexRLApp
|
15 |
Gradio app to generate remote sensing indices for properties defined by image-mask pairs.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
# IndexRLApp
|
15 |
Gradio app to generate remote sensing indices for properties defined by image-mask pairs.
|
16 |
+
|
17 |
+
## How to Install
|
18 |
+
IndexRL can be installed using the following two commands. However, it is recommended that you create a virtual environment prior to pip installation.
|
19 |
+
|
20 |
+
```cmd
|
21 |
+
pip install -r requirements.txt
|
22 |
+
python app.py
|
23 |
+
```
|
24 |
+
|
25 |
+
## Upload a Dataset
|
26 |
+
1. Set up the file structure of your dataset as follows.
|
27 |
+
|
28 |
+
```
|
29 |
+
.
|
30 |
+
βββ dataset/
|
31 |
+
βββ images/
|
32 |
+
β βββ img1.npy
|
33 |
+
β βββ img2.npy
|
34 |
+
β βββ ...
|
35 |
+
βββ masks/
|
36 |
+
βββ mask1.npy
|
37 |
+
βββ mask2.npy
|
38 |
+
βββ ...
|
39 |
+
```
|
40 |
+
|
41 |
+
2. Create a zip file for the dataset
|
42 |
+
```cmd
|
43 |
+
cd dataset
|
44 |
+
zip -r dataset.zip *
|
45 |
+
```
|
46 |
+
|
47 |
+
3. Go to the `Datasets` tab on IndexRL, upload the created zip file, give a name for the dataset, and press the `Upload` button.
|
48 |
+
|
49 |
+

|
50 |
+
|
51 |
+
## Generate Remote Sensing Indices
|
52 |
+
|
53 |
+
1. Go to the `Find Expressions` tab, select the dataset from the dropdown and click the `Find Expressions` button. With time, more and more expressions will be generated.
|
54 |
+
|
55 |
+
|
56 |
+
2. Obtain the top 5 expressions from the `Best Expressions` box.
|
57 |
+
|
58 |
+
3. Visualize and debug the expression search tree from the panels on the right.
|
59 |
+
- `Iteration` refers to each generated expression.
|
60 |
+
- `Tree Number` refers to the index of the expression tree that resulted in the selection of each symbol of the expression.
|
61 |
+
- `Expression Tree` displays the tree of symbols explored by the RL agent and the scores observed if each of those symbols were chosen. As you change the `Tree Number`, you can observe how the tree for the next choice of symbol.
|
62 |
+
|
63 |
+
|
64 |
+

|
assets/images/datasets-tab.JPG
ADDED
|
assets/images/find-expressions-tab.JPG
ADDED
|