search

WongKinYiu / yolor

implementation of paper - You Only Learn One Representation: Unified Network for Multiple Tasks (https://arxiv.org/abs/2105.04206)
GNU General Public License v3.0
1.99k stars 517 forks source link

Why we have image size as a variable in detect.py #196

Open chandralegend opened 2 years ago

fenerir commented 2 years ago

Image size affects the quality of detections. Try increasing to 1280 or 2560 to see more precise results

chandralegend commented 2 years ago

@fenerir But if the model is trained using the input size set to 1280. Don't you think that if we inference using image size set to 2560 doesn't make sense? because it will eventually get resized to 1280. Or am I missing something regarding the input CNN layer?