From 477cd84dabfd3bb63f5b5035e861cef52c0b17f5 Mon Sep 17 00:00:00 2001 From: wejoncy <247153481@qq.com> Date: Sat, 12 Oct 2024 11:19:00 +0800 Subject: [PATCH] Update README.md --- mobile/examples/object_detection/android/README.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/mobile/examples/object_detection/android/README.md b/mobile/examples/object_detection/android/README.md index c10068ca..a53eefa3 100644 --- a/mobile/examples/object_detection/android/README.md +++ b/mobile/examples/object_detection/android/README.md @@ -5,6 +5,8 @@ This is a basic Object Detection sample application for [ONNX Runtime](https://github.com/microsoft/onnxruntime) on Android with [Ort-Extensions](https://github.com/microsoft/onnxruntime-extensions) support for pre/post processing. The demo app accomplishes the task of detecting objects from a given image. The model used here is from source: [Yolov8 in extensions](https://github.com/microsoft/onnxruntime-extensions/blob/64f20828ce0291394886e277c23529cd1d11320d/tutorials/yolo_e2e.py#L37) and with pre/post processing support. +The default output model with onnxruntime-extension tools wouldn't include 'scaled_box_out_next' which is used in this example for displaying class-lable and confidence. One more step is required to get that. +Please Add a `Debug()` on top of [`onnxruntime-extensions/onnxruntime_extensions/tools/add_pre_post_processing_to_model.py:270`](https://github.com/microsoft/onnxruntime-extensions/blob/981cb049ff956a1c99ab178b36ffc83664a678f2/onnxruntime_extensions/tools/add_pre_post_processing_to_model.py#L270). This model (Yolov8n) can be fed with image bytes directly and outputs the detected objects with bounding boxes. @@ -38,4 +40,4 @@ Clone this repository to get the sample application. Then open the project under Here are some sample example screenshots of the app. App Screenshot 1 -App Screenshot 2 \ No newline at end of file +App Screenshot 2