Can YOLOv8n-seg Model Perform Real-Time Segmentation on OAK-D-Lite?

platform

Hello everyone,

I'm currently working on a project where I’m attempting to run a YOLOv8n-seg model on the OAK-D-Lite device to perform real-time segmentation. I've successfully converted the model to a blob file but encountered issues when trying to run inferences—no segmentation results appear on the OAK-D-Lite.

I would love to know if anyone has managed to get the YOLOv8n-seg model running successfully on this device. Additionally, if there are alternative segmentation models that work efficiently on the OAK-D-Lite for real-time inference, could you share your recommendations? Real-time processing is a primary goal for the project, so a model optimized for quick inferences on the Myriad X VPU would be ideal.

Thanks so much in advance for your help!

jakaskerl

platform
Check this out:
luxonis/depthai-experimentstree/master/gen2-yolo/yolo-segmentation

Thanks,
Jaka

platform

jakaskerl
Thank you for your response. Instead of processing on the host, I would like to perform the processing directly on the OAK-D-Lite. Specifically, is it possible to run the YOLOv8n-seg model independently on the Myriad X VPU?

jakaskerl

platform
The inference runs on the VPU. But there is currently not way to run decoding there as well. Decoding has to be done on host.

Thanks,
Jaka