search

cuiziteng / Illumination-Adaptive-Transformer

[BMVC 2022] You Only Need 90K Parameters to Adapt Light: A Light Weight Transformer for Image Enhancement and Exposure Correction. SOTA for low light enhancement, 0.004 seconds try this for pre-processing.
Apache License 2.0
441 stars 43 forks source link

test on 1080p images #61

Closed Yiming0416 closed 8 months ago

Yiming0416 commented 8 months ago

Hi,

Thanks for your great work and open-sourcing the code. Have you ever tried running inference on 1080p images? I am using a single RTX TITAN card with 24GB RAM and I got an CUDA out of memory error when running inference on 1080p images. Thanks.

cuiziteng commented 8 months ago

Thanks for your interest, IAT's shortage is that it occupies more cache memory, this is because of the depth-wise convolution, so maybe you could resize the image or use higher memory GPUs.