The model is trained without input normalization, so you don't need to divide 255 when testing. - Githubissues

NJU-Jet / SR_Mobile_Quantization

Winner solution of mobile AI (CVPRW 2021).

148 stars 41 forks source link

The model is trained without input normalization, so you don't need to divide 255 when testing. #14

Open NJU-Jet opened 3 years ago