HongChuang/ultralytics-ascend
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 3

ultralytics 8.0.116 NAS, DVC, YOLOv5u updates (#3124)

Browse source 
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
This commit is contained in:
Glenn Jocher 2023-06-11 20:39:32 +02:00 • committed by GitHub
parent 095b856e75
commit f4e8b39fc2
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
12 changed files with 63 additions and 41 deletions
Download patch file Download diff file Expand all files Collapse all files

19
ultralytics/yolo/utils/benchmarks.py
View file

@ -90,7 +90,7 @@ def benchmark(model=Path(SETTINGS['weights_dir']) / 'yolov8n.pt',
filename = model.ckpt_path or model.cfg
export = model # PyTorch format
else:
filename = model.export(imgsz=imgsz, format=format, half=half, int8=int8, device=device) # all others
filename = model.export(imgsz=imgsz, format=format, half=half, int8=int8, device=device, verbose=False)
export = YOLO(filename, task=model.task)
assert suffix in str(filename), 'export failed'
emoji = '❎' # indicates export succeeded
@ -196,8 +196,17 @@ class ProfileModels:
model.fuse() # to report correct params and GFLOPs in model.info()
model_info = model.info()
if self.trt and self.device.type != 'cpu' and not engine_file.is_file():
engine_file = model.export(format='engine', half=True, imgsz=self.imgsz, device=self.device)
onnx_file = model.export(format='onnx', half=True, imgsz=self.imgsz, simplify=True, device=self.device)
engine_file = model.export(format='engine',
half=True,
imgsz=self.imgsz,
device=self.device,
verbose=False)
onnx_file = model.export(format='onnx',
half=True,
imgsz=self.imgsz,
simplify=True,
device=self.device,
verbose=False)
elif file.suffix == '.onnx':
model_info = self.get_onnx_model_info(file)
onnx_file = file
@ -254,7 +263,7 @@ class ProfileModels:
for _ in range(3):
start_time = time.time()
for _ in range(self.num_warmup_runs):
model(input_data, verbose=False)
model(input_data, imgsz=self.imgsz, verbose=False)
elapsed = time.time() - start_time
# Compute number of runs as higher of min_time or num_timed_runs
@ -263,7 +272,7 @@ class ProfileModels:
# Timed runs
run_times = []
for _ in tqdm(range(num_runs), desc=engine_file):
results = model(input_data, verbose=False)
results = model(input_data, imgsz=self.imgsz, verbose=False)
run_times.append(results[0].speed['inference']) # Convert to milliseconds
run_times = self.iterative_sigma_clipping(np.array(run_times), sigma=2, max_iters=3) # sigma clipping