Mercurial > repos > bgruening > json2yolosegment

comparison yolov8.py @ 3:97bc82ee2a61 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools commit 743c8acf1ea4e4b1e718743d3772b7e592646611
author bgruening
date Mon, 14 Jul 2025 18:28:46 +0000
parents 158e6ce48345
children f6990d85161c
comparison
equal deleted inserted replaced
2:158e6ce48345 3:97bc82ee2a61
1 import argparse 1 import argparse
2 import os 2 import os
3 import pathlib 3 import pathlib
4 import shutil
5 import time 4 import time
6 from argparse import RawTextHelpFormatter 5 from argparse import RawTextHelpFormatter
7 from collections import defaultdict 6 from collections import defaultdict
8 7
9 import cv2 8 import cv2
173 172
174 173
175 # 174 #
176 # Functions 175 # Functions
177 # 176 #
178
179 def safe_rmtree(path):
180 try:
181 shutil.rmtree(path)
182 except OSError:
183 time.sleep(1)
184 shutil.rmtree(path, ignore_errors=True)
185
186
187 # Train a new model on the dataset mentioned in yaml file 177 # Train a new model on the dataset mentioned in yaml file
188 def trainModel(model_path, model_name, yaml_filepath, **kwargs): 178 def trainModel(model_path, model_name, yaml_filepath, **kwargs):
189 if "imgsz" in kwargs: 179 if "imgsz" in kwargs:
190 image_size = kwargs['imgsz'] 180 image_size = kwargs['imgsz']
191 else: 181 else:
269 if "init_lr" in kwargs: 259 if "init_lr" in kwargs:
270 init_lr = kwargs['init_lr'] 260 init_lr = kwargs['init_lr']
271 else: 261 else:
272 init_lr = 1.0 262 init_lr = 1.0
273 263
274 train_save_path = os.path.expanduser('~/runs/' + args.mode + '/train/')
275 if os.path.isdir(train_save_path):
276 safe_rmtree(train_save_path)
277 # Load a pretrained YOLO model (recommended for training) 264 # Load a pretrained YOLO model (recommended for training)
278 if args.model_format == 'pt': 265 if args.model_format == 'pt':
279 model = YOLO(os.path.join(model_path, model_name + "." + args.model_format)) 266 model = YOLO(os.path.join(model_path, model_name + "." + args.model_format))
280 else: 267 else:
281 model = YOLO(model_name + "." + args.model_format) 268 model = YOLO(model_name + "." + args.model_format)
289 return model 276 return model
290 277
291 278
292 # Validate the trained model 279 # Validate the trained model
293 def validateModel(model): 280 def validateModel(model):
294 # Remove prediction save path if already exists
295 val_save_path = os.path.expanduser('~/runs/' + args.mode + '/val/')
296 if os.path.isdir(val_save_path):
297 safe_rmtree(val_save_path)
298 # Validate the model 281 # Validate the model
299 metrics = model.val() # no args needed, dataset & settings remembered 282 metrics = model.val() # no args needed, dataset & settings remembered
300 metrics.box.map # map50-95 283 metrics.box.map # map50-95
301 metrics.box.map50 # map50 284 metrics.box.map50 # map50
302 metrics.box.map75 # map75 285 metrics.box.map75 # map75
328 if "max_det" in kwargs: 311 if "max_det" in kwargs:
329 maximum_detections = args.max_det 312 maximum_detections = args.max_det
330 else: 313 else:
331 maximum_detections = 300 314 maximum_detections = 300
332 315
333 if "run_dir" in kwargs: 316 run_save_dir = kwargs['run_dir'] # For Galaxy, run_save_dir is always provided via xml wrapper
334 run_save_dir = kwargs['run_dir']
335 else:
336 # Remove prediction save path if already exists
337 pred_save_path = os.path.expanduser('~/runs/' + args.mode + '/predict/')
338 if os.path.isdir(pred_save_path):
339 safe_rmtree(pred_save_path)
340 if "foldername" in kwargs: 317 if "foldername" in kwargs:
341 save_folder_name = kwargs['foldername'] 318 save_folder_name = kwargs['foldername']
342 # infer on a local image or directory containing images/videos 319 # infer on a local image or directory containing images/videos
343 prediction = model.predict(source=source_datapath, save=True, stream=True, 320 prediction = model.predict(source=source_datapath, save=True, stream=True,
344 conf=confidence, imgsz=image_size, 321 conf=confidence, imgsz=image_size,