TensorFlow Object Detection API Results Exporter is a Python project designed to address a specific need in the field of object detection. It extends the functionality of the Tensorflow Object Detection API by providing the capability to export images with Region of Interest (ROI) annotations and boundary box detections as XML data.
It utilizes portions of Tensorflow Object Detection API, an open-source project developed by the TensorFlow team at Google, which is licensed under the Apache License Version 2.0. I have not modified code and have only imported modules directly for the purpose of object detection.
For more information about Tensorflow Object Detection API and its license, please visit here.
Follow these steps to use the TensorFlow Object Detection API Results Exporter:
- Put your
.jpg
and.png
files into theImage
directory. - Place your Object Detection API model in the
Model
directory. - Add a
label_map.pbtxt
file to the corresponding Object Detection API model directory. - Run code in
TensorFlow Object Detection API Results Exporter.ipynb
notebook.
After completing the third step, your directory structure should resemble the following:
.
├── Image
│ ├── *.jpg
│ ├── *.png
│ └── ...
├── Model
│ ├── Object Detection API model
│ │ ├── checkpoint
│ │ │ ├── checkpoint
│ │ │ ├── ckpt-0.data-00000-of-00001
│ │ │ └── ckpt-0.index
│ │ ├── saved_model
│ │ │ ├── fingerprint.pb
│ │ │ └── saved_model.pb
│ │ ├── label_map.pbtxt
│ │ └── pipeline.config
...
After running the code in the notebook, the Image
directory will contain export_result
, object_found
, and object_not_found
subdirectories.
In the export_result
subdirectory, images with ROI and their corresponding XML annotations will be saved here.
In the object_found
subdirectory, all detected images will be moved here.
In the object_not_found
subdirectory, all images that were not detected will be moved here.
.
├── Images
│ ├── export_result
│ │ ├── *.jpg
│ │ ├── *.png
│ │ ├── *.xml
│ │ └── ...
│ ├── object_found
│ │ ├── *.jpg
│ │ ├── *.png
│ │ └── ...
│ ├── object_not_found
│ │ ├── *.jpg
│ │ ├── *.png
│ │ └── ...
...
This project is also licensed under the Apache License Version 2.0. See the LICENSE file for details.