quoccuongle/grounding-dino
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add readme for explainations

Browse Source
This commit is contained in:
SlongLiu
2023-04-12 18:11:40 +08:00
parent 049566bdc9
commit dbe0ad8f21
3 changed files with 8 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
.asset/model_explan1.PNG Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 472 KiB

BIN
.asset/model_explan2.PNG Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 456 KiB

9
README.md
View File

@@ -57,7 +57,14 @@ Marrying <a href="https://github.com/IDEA-Research/GroundingDINO">Grounding DINO
<img src="https://huggingface.co/ShilongLiu/GroundingDINO/resolve/main/GD_GLIGEN.png" alt="gd_gligen" width="100%">
</details>
## :star: Explanation/Tips for Grounding DINO Inputs and Outputs
- Grounding DINO accepts with a `(image, text)` pair as inputs.
- It will outputs `900` (by default) object boxes. Each box has a similarity scores across all input words.
- We defaultly choose the boxes whose highest similarities are higher than a `box_threshold`.
- We clip the words whose similarities are higher than the `text_threshold` as predicted labels.
- If you want to obtain objects of certain phrases, like the `dogs` in the sentence `two dogs with a stick.`, you can select the boxes with highest text similarities with `dogs` as final outputs.
![model_explain1](.asset/model_explan1.PNG)
![model_explain2](.asset/model_explan2.PNG)
## :label: TODO