quoccuongle/MultimodalOCR
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
99Franklin
2024-12-31 09:56:51 +08:00
committed by GitHub
parent c1c23501e6
commit 601bd989b5
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
OCRBench_v2/README.md
View File

@@ -51,13 +51,13 @@ Example structure of the JSON file:
```
## Evaluation Scripts
After obtaining the inference results of the model, you can see use the bellow scripts to obtain the final score of OCRBench v2. `./pred_folder/internvl2_5_26b.json` is an example inference result of InternVL2.5-26B with [VLMEvalKit](https://github.com/open-compass/VLMEvalKit). You can use `./eval_scripts/eval.py` to get the score for each samples, and the results are saved under `./res_folder`.
After obtaining the inference results from the model, you can use the following scripts to calculate the final score for OCRBench v2. For example, `./pred_folder/internvl2_5_26b.json` contains sample inference results generated by InternVL2.5-26B using [VLMEvalKit](https://github.com/open-compass/VLMEvalKit).To compute the score for each sample, you can use the script `./eval_scripts/eval.py`. The results will be saved in the `./res_folder`.
```python
python ./eval_scripts/eval.py --input_path ./pred_folder/internvl2_5_26b.json --output_path ./res_folder/internvl2_5_26b.json
```
After obtaining the scores for all samples, you can use `./eval_scripts/get_score.py` to get the metrics for OCRBench v2.
Once the scores for all samples have been calculated, you can use the script `./eval_scripts/get_score.py` to compute the overall metrics for OCRBench v2.
```python
python ./eval_scripts/get_score.py --json_file ./res_folder/internvl2_5_26b.json
@@ -78,7 +78,7 @@ python ./eval_scripts/get_score.py --json_file ./res_folder/internvl2_5_26b.json
<p>
# Copyright Statement
The data are collected from public datasets and community user contributions. This dataset is for research purposes only and not for commercial use. If there are any copyright concerns, please contact ling_fu@hust.edu.cn.
The data are collected from public datasets and community user contributions. This dataset is for research purposes only and not for commercial use. If you have any copyright concerns, please contact ling_fu@hust.edu.cn.
# Citation
Coming soon