search

594422814 / TransformerTrack

MIT License
253 stars 45 forks source link

Training setting #2

Closed chenrxi closed 3 years ago

chenrxi commented 3 years ago

How many GPUs do you use? and the size of feature map after backbone. Thank you.

594422814 commented 3 years ago

I use 4 GPUs (about 1 day to train the network). The image size is 352 \times 352, and the feature size is 22 \times 22.