When using Yolo with my CPU, it processes one minute an hour detecting people. The game tape he uses is only 20MB, while mine is 7GB. My datasets would be really, really big.
So I need a GPU. I need to do a little training datasets as well, in addition to object detection.
I'm currently considering this Acer GPU that comes as a really good price, but is 6 GB graphics card RAM enough? He uses a CPU, but uses Google Colab to train datasets. Google Colab's system RAM is 12MB, while this one's is 8MB.