I have trained a model whose encoder is resnet50 with kitti dataset.
My training command is python monodepth_main.py --mode train --model_name mono_02 --encoder resnet50 --data_path /mnt/hdd/data/kitti/ --filenames_file exps/mono_01/kitti_train_files.txt --log_directory exps/mono_02
But I found the number of trainable parameters is 58452008. It is so much larger than what you published in paper("This variant contains 48 million
trainable parameters")
Did you update the code to fix some bugs?
Is it a typo error in paper?
Hi,
I have trained a model whose encoder is resnet50 with kitti dataset. My training command is
python monodepth_main.py --mode train --model_name mono_02 --encoder resnet50 --data_path /mnt/hdd/data/kitti/ --filenames_file exps/mono_01/kitti_train_files.txt --log_directory exps/mono_02But I found the number of trainable parameters is 58452008. It is so much larger than what you published in paper("This variant contains 48 million trainable parameters")Did you update the code to fix some bugs? Is it a typo error in paper?