def train_model():
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="examples/specjale")
trainer.setTrainConfig(object_names_array=["specjal"], batch_size=4, num_experiments=10,
train_from_pretrained_model="models/yolo.h5")
trainer.trainModel()
def entrenarModelo():
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="2_RESPUESTA")
trainer.setTrainConfig(object_names_array=["RESPUESTA"],
batch_size=2,
num_experiments=50,
train_from_pretrained_model="pretrained-yolov3.h5")
trainer.trainModel()
def train_detection_model():
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="fire-dataset")
trainer.setTrainConfig(object_names_array=["fire"], batch_size=8, num_experiments=100,
train_from_pretrained_model="pretrained-yolov3.h5")
trainer.trainModel()
def train(self, data_dir, classes, batch_size, epochs):
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=data_dir)
trainer.setTrainConfig(object_names_array=classes,
batch_size=batch_size,
num_experiments=epochs,
train_from_pretrained_model=PRETRAINED_MODEL)
trainer.trainModel()
def __init__(self):
self.trainer = DetectionModelTrainer()
self.trainer.setModelTypeAsYOLOv3()
# Set name of dataset directory
self.trainer.setDataDirectory(data_directory="dog_dataset")
self.trainer.setTrainConfig(
object_names_array=['dog'],
batch_size=4,
num_experiments=10,
train_from_pretrained_model='pretrained-yolov3.h5')
def train_detection_model():
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="fire-dataset")
trainer.setTrainConfig(object_names_array=["fire"], batch_size=8, num_experiments=100,
train_from_pretrained_model="pretrained-yolov3.h5")
# download 'pretrained-yolov3.h5' from the link below
# https://github.com/OlafenwaMoses/ImageAI/releases/download/essential-v4/pretrained-yolov3.h5
trainer.trainModel()
def trainer():
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=data_directory)
trainer.setTrainConfig(
object_names_array=["seal", "no_seal"],
batch_size=4,
num_experiments=50,
train_from_pretrained_model="pretrained-yolov3.h5",
)
trainer.trainModel()
def train():
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=os.path.join('drive','MyDrive','train_data', 'circle'))
trainer.setTrainConfig(
object_names_array=['circle'],
batch_size=4,
num_experiments=200,
train_from_pretrained_model='pretrained-yolov3.h5'
)
trainer.trainModel()
def train():
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="")
trainer.setTrainConfig(
object_names_array=["Container", "Dent", "Rust", "Crack"],
batch_size=4,
num_experiments=100,
train_from_pretrained_model="pretrained-yolov3.h5",
)
trainer.trainModel()
def trainer(
data_directory="plates",
object_names_array=["LP"],
batch_size=4,
num_experiments=2,
train_from_pretrained_model="plates/models/detection_model-ex-002--loss-0007.222.h5"
):
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory)
trainer.setTrainConfig(object_names_array, batch_size, num_experiments,
train_from_pretrained_model)
trainer.trainModel()
def test_detection_training(clear_keras_session):
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=sample_dataset)
trainer.setTrainConfig(object_names_array=["hololens"], batch_size=2, num_experiments=1, train_from_pretrained_model=pretrained_model)
trainer.trainModel()
assert os.path.isdir(sample_dataset_json_folder)
assert os.path.isdir(sample_dataset_models_folder)
assert os.path.isdir(sample_dataset_cache_folder)
assert os.path.isfile(os.path.join(sample_dataset_json_folder, "detection_config.json"))
shutil.rmtree(os.path.join(sample_dataset_json_folder))
shutil.rmtree(os.path.join(sample_dataset_models_folder))
shutil.rmtree(os.path.join(sample_dataset_cache_folder))
class Trainer():
def __init__(self):
self.trainer = DetectionModelTrainer()
self.trainer.setModelTypeAsYOLOv3()
# Set name of dataset directory
self.trainer.setDataDirectory(data_directory="wheels_dataset")
self.trainer.setTrainConfig(object_names_array=['wheelchair', 'pull trolley', 'push trolley'], batch_size=8, num_experiments=10, train_from_pretrained_model='wheels-yolov3.h5')
def train(self):
self.trainer.trainModel()
def evaluate_yolov3(data_path, models_path, json_path, iou_thresh,
object_thresh, nms_thresh):
model = DetectionModelTrainer()
model.setModelTypeAsYOLOv3()
model.setDataDirectory(data_directory=data_path)
model.evaluateModel(model_path=models_path,
json_path=os.path.join(json_path,
"detection_config.json"),
iou_threshold=iou_thresh,
object_threshold=object_thresh,
nms_threshold=nms_thresh)
def eval(data_directory="plates",
model_path="plates/models",
json_path="plates/json/detection_config.json",
iou_threshold=0.5,
object_threshold=0.3,
nms_threshold=0.5):
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory)
trainer.evaluateModel(model_path, json_path, iou_threshold,
object_threshold, nms_threshold)
def keurmerk_train(data, model):
"""Trains YOLOv3 model from pretrained model.
Arguments:
data {string} -- Directory where the image and annotation data is
stored according to the specified structure.
model {string} -- Directory where the chosen model is stored.
"""
# Set model type
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
labels = [
"fairtrade",
"planetproof",
"utz",
"organic",
"vegan",
"weidemelk",
"bio",
"eko",
"asc",
"msc",
"ebio",
"beterleven",
]
# Data folder
trainer.setDataDirectory(data_directory=DATA_DIR)
# Training configuration
trainer.setTrainConfig(
object_names_array=labels,
batch_size=4,
num_experiments=100,
train_from_pretrained_model=MODEL_PATH,
)
trainer.trainModel()
def evaluarModelo():
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="2_RESPUESTA")
metrics = trainer.evaluateModel(model_path="models",
json_path="detection_config.json",
iou_threshold=0.5,
object_threshold=0.3,
nms_threshold=0.5)
print(metrics)
def train_model(self,
data_dir,
user,
objs_array,
model=None,
new_model=False):
"""
Train a specific model according with a set of data
Args:
data_dir (str): directory with the YOLO training data (pictures and annotations)
user (str): registered user which the model belongs
objs_array (list): list containing the names of unique objects to be trained for
model (str): full path of the model file to be loaded
new_model (bool): True if after training a new model file will be generated
Returns:
object: trained model
"""
if new_model:
model_path = ''
else:
self.load_model(user=user, model_path=model)
model_path = self.model_path
model_trainer = DetectionModelTrainer()
model_trainer.setModelTypeAsYOLOv3()
model_trainer.setDataDirectory(data_directory=data_dir)
model_trainer.setTrainConfig(object_names_array=list(set(objs_array)),
batch_size=4,
num_experiments=200,
train_from_pretrained_model=model_path)
model_trainer.trainModel()
if new_model:
self.version_model(user=user)
self.model = model_trainer
return self.model
def train_model(self,
model_pretrained,
label_list,
batch_size=2,
epoch=10,
use_pretrained_model=False):
self.__use_pretrained_model = use_pretrained_model
if self.__use_pretrained_model:
trainer = DetectionModelTrainer()
# fixe le type du model comme etant yoloV3
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(self.__data_dir)
trainer.setTrainConfig(
object_names_array=label_list,
batch_size=batch_size,
num_experiments=epoch,
train_from_pretrained_model=model_pretrained)
trainer.trainModel()
def main():
if (len(sys.argv) < 3):
print("Usage: " + sys.argv[0] + " <dataset_dir> <pre_trained_model>")
exit(-1)
dataset_dir = sys.argv[1]
pre_trained_model = sys.argv[2]
if (len(sys.argv) > 2):
image_name = sys.argv[2]
else:
print("Enter image or directory name: ")
image_name = input()
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=dataset_dir)
trainer.setTrainConfig(object_names_array=[
"Ripe Strawberry", "Green Strawberry", "Bad Strawberry"
],
batch_size=4,
num_experiments=40,
train_from_pretrained_model=pre_trained_model)
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="bottles")
trainer.setTrainConfig(
object_names_array=["PET", "HDPE(1)", "HDPE(2)", "PVC", "Opaque", "PP"],
batch_size=2,
num_experiments=200,
train_from_pretrained_model="pretrained-yolov3.h5")
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="ship")
# trainer.setTrainConfig(object_names_array=["hololens"], batch_size=4, num_experiments=200, train_from_pretrained_model="pretrained-yolov3.h5")
trainer.setTrainConfig(object_names_array=["ship"],
batch_size=4,
num_experiments=200,
train_from_pretrained_model="pretrained-yolov3.h5")
# In the above,when training for detecting multiple objects,
#set object_names_array=["object1", "object2", "object3",..."objectz"]
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
import os
file = "Buildings"
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="Training Data\\" + file + "\\")
trainer.setTrainConfig(object_names_array=["building"],
batch_size=1,
num_experiments=197,
train_from_pretrained_model="model\\yolo_trained.h5")
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3() # others can include RetinaNet
trainer.setDataDirectory(data_directory="inputs/hololens")
trainer.setTrainConfig(
object_names_array=["hololens"
], # an array of names of all objects in your dataset
batch_size=4,
num_experiments=100,
train_from_pretrained_model="models/pretrained-yolov3.h5"
) # use transfer learning from a pretrained YOLO model
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="Pothole")
trainer.setTrainConfig(
object_names_array=[
"Pothole Severity Low", "Pothole Severity High",
"Pothole Severity Medium"
],
batch_size=4,
num_experiments=30,
train_from_pretrained_model="pretrained-yolov3.h5"
) #download pre-trained model via https://github.com/OlafenwaMoses/ImageAI/releases/download/essential-v4/pretrained-yolov3.h5
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
import os
execution_path = os.getcwd()
trainer = DetectionModelTrainer()
trainer.setGpuUsage(1)
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=".")
trainer.setTrainConfig(object_names_array=["empty coffee", "not empty coffee"],
batch_size=4,
num_experiments=10,
train_from_pretrained_model="pretrained-yolov3.h5")
trainer.trainModel()
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="Strawberries")
trainer.setTrainConfig(
object_names_array=["red strawberry", "green strawberry"],
batch_size=4,
num_experiments=1,
train_from_pretrained_model="pretrained-yolov3.h5")
trainer.trainModel()
!pip install imageai --upgrade
#The below function is used to change the current working directory
import os
print('Previous current working directory: '+os.getcwd()+'\n')
from google.colab import drive
drive.mount('/gdrive')
#Below you should write the path to the desired directory
%cd /gdrive/My Drive/DLSC
print('\nNew current working directory: '+os.getcwd())
# Training
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="Dataset2")
trainer.setTrainConfig(object_names_array=["Vehicle"], batch_size=4, num_experiments=20, train_from_pretrained_model="pretrained-yolov3.h5")
trainer.trainModel()
# Evaluation
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="Dataset2")
metrics = trainer.evaluateModel(model_path="Dataset2/models", json_path="Dataset2/json/detection_config.json", iou_threshold=0.5, object_threshold=0.3, nms_threshold=0.5)
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="apple_dataset")
trainer.setTrainConfig(
object_names_array=["apple", "damaged_apple"],
batch_size=8,
num_experiments=50,
train_from_pretrained_model="apple_dataset\models\pretrained-yolov3.h5")
trainer.trainModel()
def train_yolo(lst_y, train_path="fs/training_yolo/", transfer_modelfile=""):
## setup
yolo = DetectionModelTrainer()
yolo.setModelTypeAsYOLOv3()
yolo.setDataDirectory(data_directory=train_path)
yolo.setTrainConfig(object_names_array=lst_y,
batch_size=4,
num_experiments=1,
train_from_pretrained_model=transfer_modelfile)
## train
print("--- training ---")
yolo.trainModel()
## evaluate
print("--- metrics ---")
metrics = yolo.evaluateModel(model_path=train_path + "models",
json_path=train_path +
"json/detection_config.json",
iou_threshold=0.5,
object_threshold=0.5,
nms_threshold=0.5)
print(metrics)
## laod model
print("--- loading model ---")
modelfile = os.listdir(train_path + "models/")[0]
return load_yolo(modelfile=train_path + "models/" + modelfile,
confjson=train_path + "/json/detection_config.json")
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory=r"G:\RYD\DataSet\new")
trainer.setTrainConfig(
object_names_array=["question"],
batch_size=4,
num_experiments=100,
train_from_pretrained_model=r"G:\ryd\yolo\pretrained-yolov3.h5")
trainer.trainModel()
