Train a YOLO classifier with 3LC metrics collection¶

Train a YOLO classifer using existing tables.

Install dependencies¶

[ ]:

%pip install 3lc
%pip install 3lc-ultralytics

Imports¶

[ ]:

import tlc
from tlc_ultralytics import YOLO, Settings

Project setup¶

[ ]:

PROJECT_NAME = "3LC Tutorials - CIFAR-10"
MODEL_NAME = "yolov8n-cls.pt"
IMAGE_COLUMN = "Image"
LABEL_COLUMN = "Label"
EPOCHS = 5
NUM_WORKERS = 0
BATCH_SIZE = 32
IMAGE_SIZE = 32
DOWNLOAR_PATH = "../../transient_data"

[ ]:

train_table = tlc.Table.from_names("initial", "CIFAR-10-train", PROJECT_NAME)
val_table = tlc.Table.from_names("initial", "CIFAR-10-val", PROJECT_NAME)

[ ]:

model = YOLO(MODEL_NAME)

settings = Settings(
    project_name=PROJECT_NAME,
    run_name="Train YOLO Classifier",
    image_embeddings_dim=2,
    conf_thres=0.2,
    sampling_weights=True,
    exclude_zero_weight_training=True,
    exclude_zero_weight_collection=False,
    image_column_name=IMAGE_COLUMN,
    label_column_name=LABEL_COLUMN,
)

model.train(
    tables={
        "train": train_table,
        "val": val_table,
    },
    settings=settings,
    batch=BATCH_SIZE,
    imgsz=IMAGE_SIZE,
    epochs=EPOCHS,
    workers=NUM_WORKERS,
    project=DOWNLOAR_PATH,
)