The main goal of this project is to create an AI easily using tensorflow
We recommend using a GPU with Hardware Acceleration for Tensorflow.
Hardware | minimum | recommended |
---|---|---|
CPU | 2 cores | 4 cores |
RAM | 4 GB | 8 GB |
GPU | 2 GB | 4 GB |
Get help:
python3 EasyAI.py --help
Launch training:
python3 EasyAI.py
Launch training without GPU:
python3 EasyAI.py --no-gpu
Launch training, save model and set dataset path:
python EasyAI.py --data_dir ./dataset/flower_photos --save test_AI.h5
Train a model with flowers dataset:
python EasyAI.py --save test_AI.h5
Predict images (class_name order is important):
python EasyAI.py --load test_AI.h5 --predict ./dataset/flower_photos/roses/ --class_name daisy dandelion roses sunflowers tulips
Command | Description | Default | Example |
---|---|---|---|
--data_dir | Path to the data directory. | ./dataset/ | --data_dir ./dataset/ |
--save-model | Save model to a HDF5 file. | None | --save test_AI.h5 |
--load-model | Load model from a HDF5 file. | None | --load test_AI.h5 |
--predict | Predict images. | None | --predict ./dataset/flower_photos/roses/ |
--class_name | Class name. | None | --class_name daisy dandelion roses sunflowers tulips |
--no-gpu | Disable GPU. | False | --no-gpu |
--batch_size | Batch size. | 32 | --batch_size 32 |
--epochs | Number of epochs. | 10 | --epochs 10 |
--model_path | Path to the model. | None | --model_path ./model/ |
--tensorboard | Enable tensorboard (Slow). | False | --tensorboard |
--checkpoint | Enable checkpoint. | False | --checkpoint |