To corret your sRGB-rendered images, you can check When Color Constancy Goes Wrong:Ĭorrecting Improperly White-Balanced Images, CVPR'19 for white balancing sRGB-rendered images (an online demo is provided).Ĭan I test images captured by camera sensors different than the camera sensors used for training (i.e., NUS, Gehler-Shi, and Cube/Cube+ datasets)? Our method works with linear raw-RGB images, not camera-rendered images. FAQ Can I use it to correct sRGB-rendered JPEG images? This is very important since all trained networks expect to get uint16 input images after applying the black/saturation level normalization. The input image file must contain the image raw-RGB values after applying the black/saturation level normalization. Each model is named based on the validation set used during the training (for example, the model trained_model_wo_CUBE+_CanonEOS550D.mat was trained using all raw-RGB linear images from NUS and Gehler-Shi datasets without including any example from the CanonEOS550D camera in Cube/Cube+ datasets). Each model was trained using different camera sensors, as discussed in our paper. You can test any of our trained models located in models directory. You can change the model_name and image_name variables to choose between our trained models and to change input image filename, respectively. The supported versions are: Matlab 2018b, Matlab 2019a, or higher.
In demo.m, you should select the version of Matlab by changing the value of Matlab_ver. Run install_.m, then run demo.m to test our trained models. The provided code for Matlab 2019b or higher gives almost the same results. The original experiments were done using Matlab 2018b. Deep Learning Toolbox for Matlab 2018b or higher.Sensor-Independent Illumination Estimation for DNN ModelsġYork University 2Samsung AI Center (SAIC) - Toronto
0 kommentar(er)
