Tensorflow mIOU and pixel accuracy bug?
Let's say I started training a tensorflow model from scratch with 1000 training steps. I get the following result at the completion of training.
Now, lets say I want to train for 2000 training steps from the previously saved checkpoint. I get the mIOU and pixel_accuracy = 1.0.
I am using '''TensorFlow v1.13.1'''. How can I fix this bug or problem ?
Solution
The problem was with my data-set. I was assigning background_tag=1, class_one=2, class_two=3. After modifying ground truth images python PIL everything worked normally.
- How to change colors of function plots in Tensorboard?
- What does Conv2D(32, (3, 3) in TensorFlow mean?
- UPOS Mappings - Tensorflow Datasets TDFS
- Obtain the output of intermediate layer (Functional API) and use it in SubClassed API
- Cannot import tensorflow in vscode
- How to use a Dense layer with an input that has a dynamically sized dimension?
- Broadcasting multiple versions of X_data that pair with the same y_data
- Low validation accuracy during data training using lfw datasets
- Deep-Learning Nan loss reasons
- How to save checkpoints for thie transformer gpt2 to continue training?
- ImportError: No module named 'matplotlib' -- Using Anaconda tensorflow environment
- How to make TensorFlow use 100% of GPU?
- using tensorflow gather for 3D tensors
- Should we do learning rate decay for adam optimizer
- What is the current «official» way/best practice to train an object detection model using the Tensorflow framework?
- module 'tensorflow._api.v1.compat.v2' has no attribute '__internal__' google colab error
- Unable to download tensorflow 2.15.0
- Questions about training LLMs on large text datasets for text generation from scratch
- Integrating numerical/physical data for CNN image classification
- Creating an ai chatbot, but getting a traceback error
- How to mask inputs with variable size in transformer model when the batches needs to be masked differently?
- Give a list of image path pairs to load using tf.data.Dataset.map but it seems to only read 2
- Compute Hessian matrix (only diagonal part) with respect to a high rank tensor
- Batch normalisation during testing
- ImportError: libcuda.so.1: cannot open shared object file
- How to have un-tracked weights in custom keras layer?
- Cannot augment images with tensorflow.keras.preprocessing.image import ImageDataGenerator
- How can I return a NumPy array using FastAPI?
- How can I convert the datatype of a numpy array sourced from an awkward array
- TensorFlow does not accept list type for dataset generator