A cucumber classification
Here, we use a cucumber image data set from Japan, which contains 9 classes of cucumber.
The details of CUCUMBER-9 is the following:
- There are 5 batches, 1400 cucumbers in each batch
- Each cucumber contains three images, including above, below and side
- The size of cucumber image is 32 x 32
- Each image have three channels (RGB)
- There are 10 classes, inculding 0 class ("others")
- Load the cucumber dataset
- Build classification model
- Train the classifier and tune hyperparameter
- Evaluate the performance of the model
- Implement an end-to-end cucumber classifier
- Realise data visualization
- ResNet 21
- Linear Classifier
- Python 3.6
- Object oriented
- TensorFlow 1.7
- TensorBoard
- TFRecord