tf_classification
tf_classification copied to clipboard
visipedia
Training, evaluation and testing code for image classification using TensorFlow
Hi, I had a question regarding the use of pretrained ImageNet model for training the classifier. The [dataset website](http://www.vision.caltech.edu/visipedia/CUB-200-2011.html) warns about using pretrained ImageNet models due to the possible overlap...
`$` CUDA_VISIBLE_DEVICES=0 python train.py \ `>` --tfrecords $DATASET_DIR/train* \ `>` --logdir $EXPERIMENT_DIR/logdir/finetune \ `>` --config $EXPERIMENT_DIR/config_train.yaml \ `>` --pretrained_model $IMAGENET_PRETRAINED_MODEL \ `>` --trainable_scopes InceptionV3/Logits InceptionV3/AuxLogits \ `>` --checkpoint_exclude_scopes InceptionV3/Logits InceptionV3/AuxLogits...
Hey, Thanks for this wonderful tutorial. I followed it with the same hyper-parameters on CUB but only got 77.4% accuracy on test data. Could you please also indicate in the...
How to turn off the data augmentation part ? Like I just want to resize the original image and train the model with it, not on the augmented images because...
Hi, thanks for this very useful code. I want to ask you some questions: - Which is the final test accuracy that you achieve? (with the method explained in the...
I would like to write a wiki about different ideas for dividing a dataset into train, val and test sets.
When starting the test script, the inception-v1 checkpoint 0 model produces 100% accuracy values for the top_k metrics. This is because the model is outputting a vector of all 0s,...
Write a tutorial for using the code to classify the CUB-200 dataset. - Data preparation - Directory structure - Configuration - Warmup training - Full training - Evaluation and Testing
The `global step` has seen a decrease with the latest changes to the input pipeline. Need to debug whether this is do the new way of doing image summaries or...