Thanks Karol for showing us around Warsaw, and of course answering my question regarding DETR's inference time! On your device it seems to be mostly around 0.2 to 0.3 s, right? I wanted to ask what you think might be the inference time on an edge device like Jetson Nano?
@@KarolMajek Hi Karol, thanks for being prompt to answer. Will wait if you can take it forward and get a Jetson nano do the inference as we might do in an edge device.
Would you mind if I ask about sharing the codes? I've already have trained model on my custom dataset, but real time video detection makes me hard. Please help me..
This wasn't realtime as I remember. Can't play it now. Check if there's OSD with FPS. colab.research.google.com/drive/1L8FpUo5SWju2zTR5jlotKZVJOvZmQbgP?usp=sharing
![Как устроен QR-код? [Veritasium]](/img/1.gif)
Sir pls make a video how to do it in code
Thanks Karol for showing us around Warsaw, and of course answering my question regarding DETR's inference time! On your device it seems to be mostly around 0.2 to 0.3 s, right? I wanted to ask what you think might be the inference time on an edge device like Jetson Nano?
Need to ty it
@@KarolMajek Hi Karol, thanks for being prompt to answer. Will wait if you can take it forward and get a Jetson nano do the inference as we might do in an edge device.
Check your thresholds. Multiple boxes are covering an object with the same size
Would you mind if I ask about sharing the codes? I've already have trained model on my custom dataset, but real time video detection makes me hard. Please help me..
This wasn't realtime as I remember. Can't play it now. Check if there's OSD with FPS.
colab.research.google.com/drive/1L8FpUo5SWju2zTR5jlotKZVJOvZmQbgP?usp=sharing
How long did it take to train? I tried to train DETR with my custom dataset, but it takes really long time
it takes a long time; this was pretrained on COCO provided by FB