Installation
pip install aek-img-trainer
For more secure way after doing that you can upgrade via:
pip install --upgrade aek-img-trainer
Usage
Create object
Now you can just use Trainer class methods example usage is shown below.
from aek_img_trainer import Trainer, Preprocessor
model = Trainer(train_root="root/trainset",
val_root="root/valset",
num_classes=16,
img_size=224,
batch_size=4,
val_reach=0.9999,
num_epochs=150,
learning_rate=1e-3,
checkpoint_path="efficientnet_b0_best_model.pth",
model_name="efficientnet_b0",
device=None,
augment=True,
scheduler=None,
scheduler_params=None,
pretrained=True)
Those hyperparameters without train and val dataset path are default if you want to use default parameters you can just give your train and val datasets' path.
model = Trainer(train_root="root/trainset",val_root="root/valset")
Training
You can train your model with parameter that created earlier.
model.train()
Prediction
You can use your model in test with below code.
model.predict(checkpoint_path="example_model.pth",
test_path="root/test.png")
Or if you give the test set path this method predict all images in the dataset folder. Your dataset should be labelled like training dataset.
Example usage:
model.predict(checkpoint_path="example_model.pth",
test_set_path="root/testset")
Save OpenVINO IR
You can save the pth model file to xml and bin files
Example usage:
model.save_openvino(
pth_path="efnet.pth",
output_path="efnet.xml",
class_number=16,
model_name="efficientnet_b0"
)
Prediction on OpenVINO IR Model
You can predict the OpenVINO IR model. Like above predict function openvino_predict function takes either test_path or test_set_path and changes the result with the given path, either predict one image or images that inside the test set. Test set should be labelled like a training dataset.
Example usage:
model.openvino_predict(
xml_path="efnet.xml",
bin_path="efnet.bin",
test_path=None,
test_set_path="root/testset"
)
Information
You can see your model's parameters and architecture.
model.print_model_info()
Help function for Trainer class
You can use help() function for get more information about functions that inside the Trainer class.
model.help()
Timm models
You can get models that inside the timm library you can use with their name in string format inside the Trainer() 'model_name' paramaters. ATTENTION: You can just use the models whose head layers are 'fc', 'head' and 'classifier'
model.list_all_timm_models()
1 maintainer
Metadata
1 maintainer
Release files for aek-img-trainer 1.0.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| aek_img_trainer-1.0.0.tar.gz | 10.0 kB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| aek_img_trainer-1.0.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 19.3 kB
Release files / aek_img_trainer-1.0.0.tar.gz
| Download URL | aek_img_trainer-1.0.0.tar.gz |
|---|---|
| Size | 10.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
a1d7175ee5c41d53cfee17cf53a367c63b96888d333b9039eca6936e6a01451c
|
|
BLAKE2b-256 checksum How to use checksums |
f72dfe2e2750ff67c268278280fbefff3c7ee0a590b8f1e4d3131b2a3fa0283e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Release files / aek_img_trainer-1.0.0-py3-none-any.whl
| Download URL | aek_img_trainer-1.0.0-py3-none-any.whl |
|---|---|
| Size | 9.4 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
0d3f84cdedfffd62085d84ed3ec0e87ac6b29fd2de4bd726ebfba10a436fa7a7
|
|
BLAKE2b-256 checksum How to use checksums |
a8a639f84dae564f90feb4d4adbcaa073dc0710cb791fb6e465cff545802cb00
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
Security for aek-img-trainer 1.0.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
