@ -0,0 +1,3 @@
# djl-image-to-text-action
Detect and classify objects in an image into texts using the SSD and ResNet models and the ImageNet dataset.
The note is not visible to the blocked user.