App splits all images and their labels using sliding window approach. Play with sliding window configuration and preview results before start splitting. All results will be saved to a new project. All crops can be mered back with another Sliding window merge app.
Industries and use cases:
- microscopic images
- satellite images
- images in huge resolution: defects detection, quality assurance in production
- split images for labelers
- split images for neural network inference
Changelog:
-
⚙️ v1.1.10 – Added new setting in sliding window configuration: sliding window size by percentage of image size, new option to make square sliding window.
-
🧹 v1.1.20 – Added support for filtering labels by a percentage of cropped area (ignore labels with less than the specified percentage of area inside crop)
| Shift window | Add padding | Change size |
|---|---|---|
 |
 |
 |
All images and their annotations will be split and saved to a new project. Also sliding window configuration is saved in project info. Just go to project -> Info tab.
Cropped image names are in the following format:
<image name>___<window index>_<window top coordinate>_<window left coordinate>.<image extension>
for example:
IMG_0748___0004_288_480.jpeg
- image name:
IMG_0748 - window index with leading zeros:
0004 - window top coordinate:
288 - window left coordinate:
480 - image extension:
jpeg
Such naming allows to perform opposite operation: merge all crops and labels to a single image another app.