xdobro23/YOLOv8-16bit
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

CoreML NMS and half fixes (#143)

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
2023-01-05 00:20:54 +01:00
committed by GitHub
parent 55bdca6768
commit 172cef2d20
18 changed files with 95 additions and 414 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
tests/check_flops.py
View File

@ -1,15 +0,0 @@
from ultralytics import YOLO
from ultralytics.yolo.utils import ROOT
if __name__ == "__main__":
for m in list((ROOT / 'yolo/v8/models').rglob('*.yaml')):
try:
YOLO(m.name, verbose=True)
except Exception as e:
print(f'ERROR for {m}: {e}')
# n vs n-seg: 8.9GFLOPs vs 12.8GFLOPs, 3.16M vs 3.6M. ch[0] // 4 (11.9GFLOPs, 3.39M)
# s vs s-seg: 28.8GFLOPs vs 44.4GFLOPs, 11.1M vs 12.9M. ch[0] // 4 (39.5GFLOPs, 11.7M)
# m vs m-seg: 79.3GFLOPs vs 113.8GFLOPs, 25.9M vs 29.5M. ch[0] // 4 (103.GFLOPs, 27.1M)
# l vs l-seg: 165.7GFLOPs vs 226.3GFLOPs, 43.7M vs 49.6M. ch[0] // 4 (207GFLOPs, 45.7M)
# x vs x-seg: 258.5GFLOPs vs 353.0GFLOPs, 68.3M vs 77.5M. ch[0] // 4 (324GFLOPs, 71.4M)

29
tests/data/dataloader/hyp_test.yaml
View File

@ -1,29 +0,0 @@
lr0: 0.001 # initial learning rate (SGD=1E-2, Adam=1E-3)
lrf: 0.01 # final OneCycleLR learning rate (lr0 * lrf)
momentum: 0.937 # SGD momentum/Adam beta1
weight_decay: 0.0005 # optimizer weight decay 5e-4
warmup_epochs: 3.0 # warmup epochs (fractions ok)
warmup_momentum: 0.8 # warmup initial momentum
warmup_bias_lr: 0.1 # warmup initial bias lr
box: 0.05 # box loss gain
cls: 0.5 # cls loss gain
cls_pw: 1.0 # cls BCELoss positive_weight
obj: 1.0 # obj loss gain (scale with pixels)
obj_pw: 1.0 # obj BCELoss positive_weight
iou_t: 0.20 # IoU training threshold
anchor_t: 4.0 # anchor-multiple threshold
# anchors: 3 # anchors per output layer (0 to ignore)
fl_gamma: 0.0 # focal loss gamma (efficientDet default gamma=1.5)
hsv_h: 0.015 # image HSV-Hue augmentation (fraction)
hsv_s: 0.7 # image HSV-Saturation augmentation (fraction)
hsv_v: 0.4 # image HSV-Value augmentation (fraction)
degrees: 0.0 # image rotation (+/- deg)
translate: 0.1 # image translation (+/- fraction)
scale: 0.5 # image scale (+/- gain)
shear: 0.0 # image shear (+/- deg)
perspective: 0.0 # image perspective (+/- fraction), range 0-0.001
flipud: 0.0 # image flip up-down (probability)
fliplr: 0.5 # image flip left-right (probability)
mosaic: 1.0 # image mosaic (probability)
mixup: 0.0 # image mixup (probability)
copy_paste: 0.5 # segment copy-paste (probability)

85
tests/data/dataloader/yolodetection.py
View File

@ -1,85 +0,0 @@
import cv2
import hydra
from ultralytics.yolo.data import build_dataloader
from ultralytics.yolo.utils import DEFAULT_CONFIG
from ultralytics.yolo.utils.plotting import plot_images
class Colors:
# Ultralytics color palette https://ultralytics.com/
def __init__(self):
# hex = matplotlib.colors.TABLEAU_COLORS.values()
hexs = ('FF3838', 'FF9D97', 'FF701F', 'FFB21D', 'CFD231', '48F90A', '92CC17', '3DDB86', '1A9334', '00D4BB',
'2C99A8', '00C2FF', '344593', '6473FF', '0018EC', '8438FF', '520085', 'CB38FF', 'FF95C8', 'FF37C7')
self.palette = [self.hex2rgb(f'#{c}') for c in hexs]
self.n = len(self.palette)
def __call__(self, i, bgr=False):
c = self.palette[int(i) % self.n]
return (c[2], c[1], c[0]) if bgr else c
@staticmethod
def hex2rgb(h): # rgb order (PIL)
return tuple(int(h[1 + i:1 + i + 2], 16) for i in (0, 2, 4))
colors = Colors() # create instance for 'from utils.plots import colors'
def plot_one_box(x, img, color=None, label=None, line_thickness=None):
import random
# Plots one bounding box on image img
tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1 # line/font thickness
color = color or [random.randint(0, 255) for _ in range(3)]
c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA)
if label:
tf = max(tl - 1, 1) # font thickness
t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0]
c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3
cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA) # filled
cv2.putText(
img,
label,
(c1[0], c1[1] - 2),
0,
tl / 3,
[225, 255, 255],
thickness=tf,
lineType=cv2.LINE_AA,
)
@hydra.main(version_base=None, config_path=str(DEFAULT_CONFIG.parent), config_name=DEFAULT_CONFIG.name)
def test(cfg):
cfg.task = "detect"
cfg.mode = "train"
dataloader, _ = build_dataloader(
cfg=cfg,
batch_size=4,
img_path="/d/dataset/COCO/coco128-seg/images",
stride=32,
label_path=None,
mode=cfg.mode,
)
for d in dataloader:
images = d["img"]
cls = d["cls"].squeeze(-1)
bboxes = d["bboxes"]
paths = d["im_file"]
batch_idx = d["batch_idx"]
result = plot_images(images, batch_idx, cls, bboxes, paths=paths)
cv2.imshow("p", result)
if cv2.waitKey(0) == ord("q"):
break
if __name__ == "__main__":
test()
# test(augment=True, rect=False)
# test(augment=False, rect=True)
# test(augment=False, rect=False)

125
tests/data/dataloader/yolopose.py
View File

@ -1,125 +0,0 @@
import cv2
import numpy as np
from omegaconf import OmegaConf
from ultralytics.yolo.data import build_dataloader
class Colors:
# Ultralytics color palette https://ultralytics.com/
def __init__(self):
# hex = matplotlib.colors.TABLEAU_COLORS.values()
hexs = ('FF3838', 'FF9D97', 'FF701F', 'FFB21D', 'CFD231', '48F90A', '92CC17', '3DDB86', '1A9334', '00D4BB',
'2C99A8', '00C2FF', '344593', '6473FF', '0018EC', '8438FF', '520085', 'CB38FF', 'FF95C8', 'FF37C7')
self.palette = [self.hex2rgb(f'#{c}') for c in hexs]
self.n = len(self.palette)
def __call__(self, i, bgr=False):
c = self.palette[int(i) % self.n]
return (c[2], c[1], c[0]) if bgr else c
@staticmethod
def hex2rgb(h): # rgb order (PIL)
return tuple(int(h[1 + i:1 + i + 2], 16) for i in (0, 2, 4))
colors = Colors() # create instance for 'from utils.plots import colors'
def plot_one_box(x, img, keypoints=None, color=None, label=None, line_thickness=None):
import random
# Plots one bounding box on image img
tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1 # line/font thickness
color = color or [random.randint(0, 255) for _ in range(3)]
c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA)
if label:
tf = max(tl - 1, 1) # font thickness
t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0]
c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3
cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA) # filled
cv2.putText(
img,
label,
(c1[0], c1[1] - 2),
0,
tl / 3,
[225, 255, 255],
thickness=tf,
lineType=cv2.LINE_AA,
)
if keypoints is not None:
plot_keypoint(img, keypoints, color, tl)
def plot_keypoint(img, keypoints, color, tl):
num_l = len(keypoints)
# clors = [(255, 0, 0),(0, 255, 0),(0, 0, 255),(255, 255, 0),(0, 255, 255)]
# clors = [[random.randint(0, 255) for _ in range(3)] for _ in range(num_l)]
for i in range(num_l):
point_x = int(keypoints[i][0])
point_y = int(keypoints[i][1])
cv2.circle(img, (point_x, point_y), tl + 3, color, -1)
with open("ultralytics/tests/data/dataloader/hyp_test.yaml") as f:
hyp = OmegaConf.load(f)
def test(augment, rect):
dataloader, _ = build_dataloader(
img_path="/d/dataset/COCO/images/val2017",
imgsz=640,
label_path=None,
cache=False,
hyp=hyp,
augment=augment,
prefix="",
rect=rect,
batch_size=4,
stride=32,
pad=0.5,
use_segments=False,
use_keypoints=True,
)
for d in dataloader:
idx = 1 # show which image inside one batch
img = d["img"][idx].numpy()
img = np.ascontiguousarray(img.transpose(1, 2, 0))
ih, iw = img.shape[:2]
# print(img.shape)
bidx = d["batch_idx"]
cls = d["cls"][bidx == idx].numpy()
bboxes = d["bboxes"][bidx == idx].numpy()
bboxes[:, [0, 2]] *= iw
bboxes[:, [1, 3]] *= ih
keypoints = d["keypoints"][bidx == idx]
keypoints[..., 0] *= iw
keypoints[..., 1] *= ih
# print(keypoints, keypoints.shape)
# print(d["im_file"])
for i, b in enumerate(bboxes):
x, y, w, h = b
x1 = x - w / 2
x2 = x + w / 2
y1 = y - h / 2
y2 = y + h / 2
c = int(cls[i][0])
# print(x1, y1, x2, y2)
plot_one_box([int(x1), int(y1), int(x2), int(y2)],
img,
keypoints=keypoints[i],
label=f"{c}",
color=colors(c))
cv2.imshow("p", img)
if cv2.waitKey(0) == ord("q"):
break
if __name__ == "__main__":
test(augment=True, rect=False)
test(augment=False, rect=True)
test(augment=False, rect=False)

85
tests/data/dataloader/yolosegment.py
View File

@ -1,85 +0,0 @@
import cv2
import hydra
from ultralytics.yolo.data import build_dataloader
from ultralytics.yolo.utils import DEFAULT_CONFIG
from ultralytics.yolo.utils.plotting import plot_images
class Colors:
# Ultralytics color palette https://ultralytics.com/
def __init__(self):
# hex = matplotlib.colors.TABLEAU_COLORS.values()
hexs = ('FF3838', 'FF9D97', 'FF701F', 'FFB21D', 'CFD231', '48F90A', '92CC17', '3DDB86', '1A9334', '00D4BB',
'2C99A8', '00C2FF', '344593', '6473FF', '0018EC', '8438FF', '520085', 'CB38FF', 'FF95C8', 'FF37C7')
self.palette = [self.hex2rgb(f'#{c}') for c in hexs]
self.n = len(self.palette)
def __call__(self, i, bgr=False):
c = self.palette[int(i) % self.n]
return (c[2], c[1], c[0]) if bgr else c
@staticmethod
def hex2rgb(h): # rgb order (PIL)
return tuple(int(h[1 + i:1 + i + 2], 16) for i in (0, 2, 4))
colors = Colors() # create instance for 'from utils.plots import colors'
def plot_one_box(x, img, color=None, label=None, line_thickness=None):
import random
# Plots one bounding box on image img
tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1 # line/font thickness
color = color or [random.randint(0, 255) for _ in range(3)]
c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA)
if label:
tf = max(tl - 1, 1) # font thickness
t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0]
c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3
cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA) # filled
cv2.putText(
img,
label,
(c1[0], c1[1] - 2),
0,
tl / 3,
[225, 255, 255],
thickness=tf,
lineType=cv2.LINE_AA,
)
@hydra.main(version_base=None, config_path=str(DEFAULT_CONFIG.parent), config_name=DEFAULT_CONFIG.name)
def test(cfg):
cfg.task = "segment"
cfg.mode = "train"
dataloader, _ = build_dataloader(
cfg=cfg,
batch_size=4,
img_path="/d/dataset/COCO/coco128-seg/images",
stride=32,
label_path=None,
mode=cfg.mode,
)
for d in dataloader:
images = d["img"]
masks = d["masks"]
cls = d["cls"].squeeze(-1)
bboxes = d["bboxes"]
paths = d["im_file"]
batch_idx = d["batch_idx"]
result = plot_images(images, batch_idx, cls, bboxes, masks, paths=paths)
cv2.imshow("p", result)
if cv2.waitKey(0) == ord("q"):
break
if __name__ == "__main__":
test()
# test(augment=True, rect=False)
# test(augment=False, rect=True)
# test(augment=False, rect=False)

32
tests/functional/test_loaders.py
View File

@ -1,32 +0,0 @@
from ultralytics.nn.tasks import DetectionModel
from ultralytics.yolo.utils.checks import check_yaml
def test_model_parser():
cfg = check_yaml("yolov8n.yaml") # check YAML
# Create model
model = DetectionModel(cfg)
model.info()
'''
# Options
if opt.line_profile: # profile layer by layer
model(im, profile=True)
elif opt.profile: # profile forward-backward
results = profile(input=im, ops=[model], n=3)
elif opt.test: # test all models
for cfg in Path(ROOT / 'models').rglob('yolo*.yaml'):
try:
_ = Model(cfg)
except Exception as e:
print(f'Error in {cfg}: {e}')
else: # report fused model summary
model.fuse()
'''
if __name__ == "__main__":
test_model_parser()

14
tests/tests_cli.py → tests/test_cli.py
View File

@ -1,6 +1,10 @@
import os
from pathlib import Path
from ultralytics.yolo.utils import ROOT
from ultralytics.yolo.utils import ROOT, SETTINGS
MODEL = Path(SETTINGS['weights_dir']) / 'yolov8n.pt'
CFG = 'yolov8n.yaml'
def test_checks():
@ -9,7 +13,7 @@ def test_checks():
# Train checks ---------------------------------------------------------------------------------------------------------
def test_train_detect():
os.system('yolo mode=train task=detect model=yolov8n.yaml data=coco128.yaml imgsz=32 epochs=1')
os.system(f'yolo mode=train task=detect model={MODEL} data=coco128.yaml imgsz=32 epochs=1')
def test_train_segment():
@ -22,7 +26,7 @@ def test_train_classify():
# Val checks -----------------------------------------------------------------------------------------------------------
def test_val_detect():
os.system('yolo mode=val task=detect model=yolov8n.pt data=coco128.yaml imgsz=32 epochs=1')
os.system(f'yolo mode=val task=detect model={MODEL} data=coco128.yaml imgsz=32 epochs=1')
def test_val_segment():
@ -35,7 +39,7 @@ def test_val_classify():
# Predict checks -------------------------------------------------------------------------------------------------------
def test_predict_detect():
os.system(f"yolo mode=predict model=yolov8n.pt source={ROOT / 'assets'}")
os.system(f"yolo mode=predict model={MODEL} source={ROOT / 'assets'}")
def test_predict_segment():
@ -48,7 +52,7 @@ def test_predict_classify():
# Export checks --------------------------------------------------------------------------------------------------------
def test_export_detect_torchscript():
os.system('yolo mode=export model=yolov8n.pt format=torchscript')
os.system(f'yolo mode=export model={MODEL} format=torchscript')
def test_export_segment_torchscript():

11
tests/tests.py → tests/test_python.py
View File

@ -1,9 +1,11 @@
from pathlib import Path
import torch
from ultralytics import YOLO
from ultralytics.yolo.utils import ROOT
from ultralytics.yolo.utils import ROOT, SETTINGS
MODEL = ROOT / 'weights/yolov8n.pt'
MODEL = Path(SETTINGS['weights_dir']) / 'yolov8n.pt'
CFG = 'yolov8n.yaml'
@ -95,6 +97,11 @@ def test_export_paddle():
model.export(format='paddle')
def test_all_model_yamls():
for m in list((ROOT / 'yolo/v8/models').rglob('*.yaml')):
YOLO(m.name)
# def run_all_tests(): # do not name function test_...
# pass
#