Does NXP have an example of object detection for iMXRT700 that run on CPU/NPU? I have a question that is same as the title. I have an object detection tasks, and I want to deploy on i.MXRT700. Before I have seen the example with tflm_cifar or tflm_label_image. It is good example to follow and there is a guideline to switch the classification model with NPU or CPU. An It also supports static image and from the camera. The model was the tflite model. Now with object detection task, do you have an example for object detection task? I am interesting about it, and it is good if there is an example for object detection that support: - NPU/CPU runntime - Static image / webcam - tflite or glow These features are same as in the tflm_cifar or tflm_label. I am looking forward for your support. Thank you. Re: Does NXP have an example of object detection for iMXRT700 that run on CPU/NPU? Hi @nnxxpp ,
Thank you so much for your interest in our products and for using our community.
For your object detection application on the i.MX RT700, I would recommend referring to the following documents for more information about Vision AI/ML applications and the object detection example:
1: Vision AI/ML with i.MX RT700
2: AN14718: USB Camera Object Detection Pipeline on i.MX RT700
I hope these resources will be helpful for your development.
Best Regards
May Liu Re: Does NXP have an example of object detection for iMXRT700 that run on CPU/NPU? @mayliu1 Thank you for your support. For your object detection application on the i.MX RT700, I would recommend referring to the following documents for more information about Vision AI/ML applications and the object detection example: 1: Vision AI/ML with i.MX RT700 2: AN14718:USB Camera Object Detection Pipeline on i.MX RT700 I have read these 2 items for i.MX RT700. It is for USB camera and specific LCD. I saw that, it mentions about source code for this example, but I did not see where I could download this source code. I also checked in MCUXpresso IDE 25.6, but in the eiq examples, there is not this one. There is not the example in the Application Hub https://mcuxpresso.nxp.com/appcodehub?category=vision Could you please give me a link to download source code for this example? Currently I am using MCUXpresso IDE. Moreover, I have question, in this example: - It uses USB camera, it means that, I can use any USB camera. There is not limitation. Is that right? - About LCD, I do not have the LCD that is mentioned on the guide. Could I modify the source code to run without LCD and print results on the my PC via Serial port (same as tflm_cifar classification example)? - In this example, the model is FastDet, it is converted to tflite model. Does NXP have an example with ONNX/GLow model? - FastDet model here is trained with Pytorch in the original version https://github.com/dog-qiuqiu/FastestDet . But NXP released tflite FastestDet. I think that NXP has procedure and example to get tflite FastestDet from the original repo. Does NXP share it? Because with custom dataset, we need to finetune model and do many things to get tflite FastestDet. If not, it is difficult to reproduce results and to use new models. Thank you.
記事全体を表示