video ccn2d resnet-152 (pretr imagenet) 224x224 2018-08-10 05:54:51,629:CRITICAL:EmoV2_step40: iteration: 16000: Loss: 0.366158, lr: 0.008417 2018-08-10 06:02:58,199:CRITICAL:EmoV2_step40: validate. Iteration: 16000: Accuracy (valence, arousal): 40.265% 20.354% 2018-08-10 06:02:58,200:CRITICAL:EmoV2_step40: validate. Iteration: 16000: Loss: 0.998851
video ccn3d resnet-18 (pretr kinet) 128x128x12 2018-08-11 06:55:55,034:CRITICAL:EmoV2_step40: iteration: 78000: Loss: 0.567958, lr: 0.002685 Valence: Accuracy3D per clips: 37.611% error per clips: mean=0.161, std=0.106 Arousal: Accuracy3D per clips: 21.681% error per clips: mean=0.263, std=0.225
2018-08-11 06:56:00,042:CRITICAL:EmoV2_step40: validate. Iteration: 78000: Accuracy (valence, arousal): 37.611% 21.681%