I'm wondering if LFFD would be a good candidate to track the basketball in the context of a game. As you can imagine, there are a lot of partial occlusions, and it moves quite fast. Of course, would be a custom trained model, as it is something not supported at this stage, but enters the "single category" requirement
Secondly, I am aware of the OpenVINO port of LFFD. Do you know if it supports Movidius Myriad X VPU? Anybody has measured the performance of such implementation, especially with a board attached to the PCI slot instead of the USB?
Hello Yonghao,
I'm wondering if LFFD would be a good candidate to track the basketball in the context of a game. As you can imagine, there are a lot of partial occlusions, and it moves quite fast. Of course, would be a custom trained model, as it is something not supported at this stage, but enters the "single category" requirement
Secondly, I am aware of the OpenVINO port of LFFD. Do you know if it supports Movidius Myriad X VPU? Anybody has measured the performance of such implementation, especially with a board attached to the PCI slot instead of the USB?
Thank you