ultralytics 8.1.21 Add YOLOv8-World-v2 models (#8580)
Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com> Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com> Co-authored-by: UltralyticsAssistant <web@ultralytics.com>
This commit is contained in:
Laughing
GitHub
parent
906b8d31dc
commit
946e18f79c
12 changed files with 98 additions and 48 deletions
|
|
@ -32,7 +32,7 @@ from pathlib import Path
|
|||
import numpy as np
|
||||
import torch.cuda
|
||||
|
||||
from ultralytics import YOLO
|
||||
from ultralytics import YOLO, YOLOWorld
|
||||
from ultralytics.cfg import TASK2DATA, TASK2METRIC
|
||||
from ultralytics.engine.exporter import export_formats
|
||||
from ultralytics.utils import ASSETS, LINUX, LOGGER, MACOS, TQDM, WEIGHTS_DIR
|
||||
|
|
@ -84,14 +84,20 @@ def benchmark(
|
|||
emoji, filename = "❌", None # export defaults
|
||||
try:
|
||||
# Checks
|
||||
if i == 9:
|
||||
if i == 9: # Edge TPU
|
||||
assert LINUX, "Edge TPU export only supported on Linux"
|
||||
elif i == 7:
|
||||
elif i == 7: # TF GraphDef
|
||||
assert model.task != "obb", "TensorFlow GraphDef not supported for OBB task"
|
||||
elif i in {5, 10}: # CoreML and TF.js
|
||||
assert MACOS or LINUX, "export only supported on macOS and Linux"
|
||||
if i in {3, 5}: # CoreML and OpenVINO
|
||||
assert not IS_PYTHON_3_12, "CoreML and OpenVINO not supported on Python 3.12"
|
||||
if i in {6, 7, 8, 9, 10}: # All TF formats
|
||||
assert not isinstance(model, YOLOWorld), "YOLOWorldv2 TensorFlow exports not supported by onnx2tf yet"
|
||||
if i in {11}: # Paddle
|
||||
assert not isinstance(model, YOLOWorld), "YOLOWorldv2 Paddle exports not supported yet"
|
||||
if i in {12}: # NCNN
|
||||
assert not isinstance(model, YOLOWorld), "YOLOWorldv2 NCNN exports not supported yet"
|
||||
if "cpu" in device.type:
|
||||
assert cpu, "inference not supported on CPU"
|
||||
if "cuda" in device.type:
|
||||
|
|
@ -261,7 +267,8 @@ class ProfileModels:
|
|||
"""
|
||||
return 0.0, 0.0, 0.0, 0.0 # return (num_layers, num_params, num_gradients, num_flops)
|
||||
|
||||
def iterative_sigma_clipping(self, data, sigma=2, max_iters=3):
|
||||
@staticmethod
|
||||
def iterative_sigma_clipping(data, sigma=2, max_iters=3):
|
||||
"""Applies an iterative sigma clipping algorithm to the given data times number of iterations."""
|
||||
data = np.array(data)
|
||||
for _ in range(max_iters):
|
||||
|
|
@ -359,9 +366,13 @@ class ProfileModels:
|
|||
def generate_table_row(self, model_name, t_onnx, t_engine, model_info):
|
||||
"""Generates a formatted string for a table row that includes model performance and metric details."""
|
||||
layers, params, gradients, flops = model_info
|
||||
return f"| {model_name:18s} | {self.imgsz} | - | {t_onnx[0]:.2f} ± {t_onnx[1]:.2f} ms | {t_engine[0]:.2f} ± {t_engine[1]:.2f} ms | {params / 1e6:.1f} | {flops:.1f} |"
|
||||
return (
|
||||
f"| {model_name:18s} | {self.imgsz} | - | {t_onnx[0]:.2f} ± {t_onnx[1]:.2f} ms | {t_engine[0]:.2f} ± "
|
||||
f"{t_engine[1]:.2f} ms | {params / 1e6:.1f} | {flops:.1f} |"
|
||||
)
|
||||
|
||||
def generate_results_dict(self, model_name, t_onnx, t_engine, model_info):
|
||||
@staticmethod
|
||||
def generate_results_dict(model_name, t_onnx, t_engine, model_info):
|
||||
"""Generates a dictionary of model details including name, parameters, GFLOPS and speed metrics."""
|
||||
layers, params, gradients, flops = model_info
|
||||
return {
|
||||
|
|
@ -372,11 +383,18 @@ class ProfileModels:
|
|||
"model/speed_TensorRT(ms)": round(t_engine[0], 3),
|
||||
}
|
||||
|
||||
def print_table(self, table_rows):
|
||||
@staticmethod
|
||||
def print_table(table_rows):
|
||||
"""Formats and prints a comparison table for different models with given statistics and performance data."""
|
||||
gpu = torch.cuda.get_device_name(0) if torch.cuda.is_available() else "GPU"
|
||||
header = f"| Model | size<br><sup>(pixels) | mAP<sup>val<br>50-95 | Speed<br><sup>CPU ONNX<br>(ms) | Speed<br><sup>{gpu} TensorRT<br>(ms) | params<br><sup>(M) | FLOPs<br><sup>(B) |"
|
||||
separator = "|-------------|---------------------|--------------------|------------------------------|-----------------------------------|------------------|-----------------|"
|
||||
header = (
|
||||
f"| Model | size<br><sup>(pixels) | mAP<sup>val<br>50-95 | Speed<br><sup>CPU ONNX<br>(ms) | "
|
||||
f"Speed<br><sup>{gpu} TensorRT<br>(ms) | params<br><sup>(M) | FLOPs<br><sup>(B) |"
|
||||
)
|
||||
separator = (
|
||||
"|-------------|---------------------|--------------------|------------------------------|"
|
||||
"-----------------------------------|------------------|-----------------|"
|
||||
)
|
||||
|
||||
print(f"\n\n{header}")
|
||||
print(separator)
|
||||
|
|
|
|||
|
|
@ -20,7 +20,8 @@ GITHUB_ASSETS_NAMES = (
|
|||
[f"yolov8{k}{suffix}.pt" for k in "nsmlx" for suffix in ("", "-cls", "-seg", "-pose", "-obb")]
|
||||
+ [f"yolov5{k}{resolution}u.pt" for k in "nsmlx" for resolution in ("", "6")]
|
||||
+ [f"yolov3{k}u.pt" for k in ("", "-spp", "-tiny")]
|
||||
+ [f"yolov8{k}-world.pt" for k in "sml"]
|
||||
+ [f"yolov8{k}-world.pt" for k in "smlx"]
|
||||
+ [f"yolov8{k}-worldv2.pt" for k in "smlx"]
|
||||
+ [f"yolo_nas_{k}.pt" for k in "sml"]
|
||||
+ [f"sam_{k}.pt" for k in "bl"]
|
||||
+ [f"FastSAM-{k}.pt" for k in "sx"]
|
||||
|
|
|
|||
|
|
@ -60,27 +60,29 @@ def imshow(winname: str, mat: np.ndarray):
|
|||
_torch_save = torch.save # copy to avoid recursion errors
|
||||
|
||||
|
||||
def torch_save(*args, **kwargs):
|
||||
def torch_save(*args, use_dill=True, **kwargs):
|
||||
"""
|
||||
Use dill (if exists) to serialize the lambda functions where pickle does not do this. Also adds 3 retries with
|
||||
exponential standoff in case of save failure to improve robustness to transient issues.
|
||||
Optionally use dill to serialize lambda functions where pickle does not, adding robustness with 3 retries and
|
||||
exponential standoff in case of save failure.
|
||||
|
||||
Args:
|
||||
*args (tuple): Positional arguments to pass to torch.save.
|
||||
use_dill (bool): Whether to try using dill for serialization if available. Defaults to True.
|
||||
**kwargs (dict): Keyword arguments to pass to torch.save.
|
||||
"""
|
||||
try:
|
||||
import dill as pickle # noqa
|
||||
except ImportError:
|
||||
assert use_dill
|
||||
import dill as pickle
|
||||
except (AssertionError, ImportError):
|
||||
import pickle
|
||||
|
||||
if "pickle_module" not in kwargs:
|
||||
kwargs["pickle_module"] = pickle # noqa
|
||||
kwargs["pickle_module"] = pickle
|
||||
|
||||
for i in range(4): # 3 retries
|
||||
try:
|
||||
return _torch_save(*args, **kwargs)
|
||||
except RuntimeError: # unable to save, possibly waiting for device to flush or anti-virus to finish scanning
|
||||
except RuntimeError as e: # unable to save, possibly waiting for device to flush or antivirus scan
|
||||
if i == 3:
|
||||
raise
|
||||
time.sleep((2**i) / 2) # exponential standoff 0.5s, 1.0s, 2.0s
|
||||
raise e
|
||||
time.sleep((2**i) / 2) # exponential standoff: 0.5s, 1.0s, 2.0s
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue