Bienqq
/
car-detection-bayes
2
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity

updates

This commit is contained in:
Glenn Jocher 2019-10-08 13:25:50 +02:00
parent cfc562c2c8
commit a18ad6025f
1 changed files with 144 additions and 64 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

178
utils/datasets.py
View File

@ -99,7 +99,7 @@ class LoadImages: # for inference
print('image %g/%g %s: ' % (self.count, self.nF, path), end='') print('image %g/%g %s: ' % (self.count, self.nF, path), end='')
# Padded resize # Padded resize
img, *_ = letterbox(img0, new_shape=self.img_size) img = letterbox(img0, new_shape=self.img_size)[0]
# Normalize RGB # Normalize RGB
img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB
@ -172,7 +172,7 @@ class LoadWebcam: # for inference
print('webcam %g: ' % self.count, end='') print('webcam %g: ' % self.count, end='')
# Padded resize # Padded resize
img, *_ = letterbox(img0, new_shape=self.img_size) img = letterbox(img0, new_shape=self.img_size)[0]
# Normalize RGB # Normalize RGB
img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB
@ -406,41 +406,22 @@ class LoadImagesAndLabels(Dataset): # for training/testing
label_path = self.label_files[index] label_path = self.label_files[index]
hyp = self.hyp hyp = self.hyp
# Load image mosaic = True # load 4 images at a time into a mosaic
img = self.imgs[index] if mosaic:
if img is None: # Load mosaic
img = cv2.imread(img_path) # BGR img, labels = load_mosaic(self, index)
assert img is not None, 'Image Not Found ' + img_path
r = self.img_size / max(img.shape) # size ratio
if self.augment and r < 1: # if training (NOT testing), downsize to inference shape
h, w, _ = img.shape h, w, _ = img.shape
img = cv2.resize(img, (int(w * r), int(h * r)), interpolation=cv2.INTER_LINEAR) # INTER_LINEAR fastest
# Augment colorspace else:
augment_hsv = True # Load image
if self.augment and augment_hsv: img = load_image(self, index)
# SV augmentation by 50%
img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV) # hue, sat, val
S = img_hsv[:, :, 1].astype(np.float32) # saturation
V = img_hsv[:, :, 2].astype(np.float32) # value
a = random.uniform(-1, 1) * hyp['hsv_s'] + 1
b = random.uniform(-1, 1) * hyp['hsv_v'] + 1
S *= a
V *= b
img_hsv[:, :, 1] = S if a < 1 else S.clip(None, 255)
img_hsv[:, :, 2] = V if b < 1 else V.clip(None, 255)
cv2.cvtColor(img_hsv, cv2.COLOR_HSV2BGR, dst=img)
# Letterbox # Letterbox
h, w, _ = img.shape h, w, _ = img.shape
if self.rect: if self.rect:
shape = self.batch_shapes[self.batch[index]] img, ratio, padw, padh = letterbox(img, self.batch_shapes[self.batch[index]], mode='rect')
img, ratiow, ratioh, padw, padh = letterbox(img, new_shape=shape, mode='rect')
else: else:
shape = self.img_size img, ratio, padw, padh = letterbox(img, self.img_size, mode='square')
img, ratiow, ratioh, padw, padh = letterbox(img, new_shape=shape, mode='square')
# Load labels # Load labels
labels = [] labels = []
@ -453,22 +434,26 @@ class LoadImagesAndLabels(Dataset): # for training/testing
if x.size > 0: if x.size > 0:
# Normalized xywh to pixel xyxy format # Normalized xywh to pixel xyxy format
labels = x.copy() labels = x.copy()
labels[:, 1] = ratiow * w * (x[:, 1] - x[:, 3] / 2) + padw labels[:, 1] = ratio[0] * w * (x[:, 1] - x[:, 3] / 2) + padw
labels[:, 2] = ratioh * h * (x[:, 2] - x[:, 4] / 2) + padh labels[:, 2] = ratio[1] * h * (x[:, 2] - x[:, 4] / 2) + padh
labels[:, 3] = ratiow * w * (x[:, 1] + x[:, 3] / 2) + padw labels[:, 3] = ratio[0] * w * (x[:, 1] + x[:, 3] / 2) + padw
labels[:, 4] = ratioh * h * (x[:, 2] + x[:, 4] / 2) + padh labels[:, 4] = ratio[1] * h * (x[:, 2] + x[:, 4] / 2) + padh
# Augment image and labels
if self.augment: if self.augment:
img, labels = random_affine(img, labels, # Augment colorspace
degrees=hyp['degrees'], augment_hsv(img, hgain=hyp['hsv_h'], sgain=hyp['hsv_s'], vgain=0.0)
translate=hyp['translate'],
scale=hyp['scale'],
shear=hyp['shear'])
# Cutout # Augment imagespace
if random.random() < 0.9: g = 0.0 if mosaic else 1.0 # do not augment mosaics
labels = cutout(img, labels) img, labels = random_affine(img, labels,
degrees=hyp['degrees'] * g,
translate=hyp['translate'] * g,
scale=hyp['scale'] * g,
shear=hyp['shear'] * g)
# Apply cutouts
# if random.random() < 0.9:
# labels = cutout(img, labels)
nL = len(labels) # number of labels nL = len(labels) # number of labels
if nL: if nL:
@ -513,17 +498,112 @@ class LoadImagesAndLabels(Dataset): # for training/testing
return torch.stack(img, 0), torch.cat(label, 0), path, hw return torch.stack(img, 0), torch.cat(label, 0), path, hw
def load_image(self, index):
# loads 1 image from dataset
img = self.imgs[index]
if img is None:
img_path = self.img_files[index]
img = cv2.imread(img_path) # BGR
assert img is not None, 'Image Not Found ' + img_path
r = self.img_size / max(img.shape) # size ratio
if self.augment and r < 1: # if training (NOT testing), downsize to inference shape
h, w, _ = img.shape
img = cv2.resize(img, (int(w * r), int(h * r)), interpolation=cv2.INTER_LINEAR) # _LINEAR fastest
return img
def augment_hsv(img, hgain=0.5, sgain=0.5, vgain=0.5):
# SV augmentation by 50%
img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV) # hue, sat, val
S = img_hsv[:, :, 1].astype(np.float32) # saturation
V = img_hsv[:, :, 2].astype(np.float32) # value
a = random.uniform(-1, 1) * sgain + 1
b = random.uniform(-1, 1) * vgain + 1
S *= a
V *= b
img_hsv[:, :, 1] = S if a < 1 else S.clip(None, 255)
img_hsv[:, :, 2] = V if b < 1 else V.clip(None, 255)
cv2.cvtColor(img_hsv, cv2.COLOR_HSV2BGR, dst=img) # no return needed
def load_mosaic(self, index):
# loads up images in a mosaic
labels4 = []
s = self.img_size
xc, yc = [int(random.uniform(s * 0.5, s * 1.5)) for _ in range(2)] # mosaic center x, y
img4 = np.zeros((s * 2, s * 2, 3), dtype=np.uint8) + 128 # base image with 4 tiles
indices = [index] + [random.randint(0, len(self.labels) - 1) for _ in range(3)] # 3 additional image indices
for i, index in enumerate(indices):
# Load image
img = load_image(self, index)
h, w, _ = img.shape
# merge img into img4
if i == 0: # top left
x1a, y1a, x2a, y2a = max(xc - w, 0), max(yc - h, 0), xc, yc # xmin, ymin, xmax, ymax (large image)
x1b, y1b, x2b, y2b = w - (x2a - x1a), h - (y2a - y1a), w, h # xmin, ymin, xmax, ymax (small image)
elif i == 1: # top right
x1a, y1a, x2a, y2a = xc, max(yc - h, 0), min(xc + w, s * 2), yc
x1b, y1b, x2b, y2b = 0, h - (y2a - y1a), min(w, x2a - x1a), h
elif i == 2: # bottom left
x1a, y1a, x2a, y2a = max(xc - w, 0), yc, xc, min(s * 2, yc + h)
x1b, y1b, x2b, y2b = w - (x2a - x1a), 0, max(xc, w), min(y2a - y1a, h)
elif i == 3: # bottom right
x1a, y1a, x2a, y2a = xc, yc, min(xc + w, s * 2), min(s * 2, yc + h)
x1b, y1b, x2b, y2b = 0, 0, min(w, x2a - x1a), min(y2a - y1a, h)
img4[y1a:y2a, x1a:x2a] = img[y1b:y2b, x1b:x2b] # img4[ymin:ymax, xmin:xmax]
padw = x1a - x1b
padh = y1a - y1b
# Load labels
label_path = self.label_files[index]
if os.path.isfile(label_path):
x = self.labels[index]
if x is None: # labels not preloaded
with open(label_path, 'r') as f:
x = np.array([x.split() for x in f.read().splitlines()], dtype=np.float32)
if x.size > 0:
# Normalized xywh to pixel xyxy format
labels = x.copy()
labels[:, 1] = w * (x[:, 1] - x[:, 3] / 2) + padw
labels[:, 2] = h * (x[:, 2] - x[:, 4] / 2) + padh
labels[:, 3] = w * (x[:, 1] + x[:, 3] / 2) + padw
labels[:, 4] = h * (x[:, 2] + x[:, 4] / 2) + padh
labels4.append(labels)
labels4 = np.concatenate(labels4, 0)
# hyp = self.hyp
# img4, labels4 = random_affine(img4, labels4,
# degrees=hyp['degrees'],
# translate=hyp['translate'],
# scale=hyp['scale'],
# shear=hyp['shear'])
# Center crop
a = s // 2
img4 = img4[a:a + s, a:a + s]
labels4[:, 1:] -= a
return img4, labels4
def letterbox(img, new_shape=416, color=(128, 128, 128), mode='auto', interp=cv2.INTER_AREA): def letterbox(img, new_shape=416, color=(128, 128, 128), mode='auto', interp=cv2.INTER_AREA):
# Resize a rectangular image to a 32 pixel multiple rectangle # Resize a rectangular image to a 32 pixel multiple rectangle
# https://github.com/ultralytics/yolov3/issues/232 # https://github.com/ultralytics/yolov3/issues/232
shape = img.shape[:2] # current shape [height, width] shape = img.shape[:2] # current shape [height, width]
if isinstance(new_shape, int): if isinstance(new_shape, int):
ratio = float(new_shape) / max(shape) r = float(new_shape) / max(shape) # ratio = new / old
else: else:
ratio = max(new_shape) / max(shape) # ratio = new / old r = max(new_shape) / max(shape)
ratiow, ratioh = ratio, ratio ratio = r, r # width, height ratios
new_unpad = (int(round(shape[1] * ratio)), int(round(shape[0] * ratio))) new_unpad = (int(round(shape[1] * r)), int(round(shape[0] * r)))
# Compute padding https://github.com/ultralytics/yolov3/issues/232 # Compute padding https://github.com/ultralytics/yolov3/issues/232
if mode is 'auto': # minimum rectangle if mode is 'auto': # minimum rectangle
@ -538,14 +618,14 @@ def letterbox(img, new_shape=416, color=(128, 128, 128), mode='auto', interp=cv2
elif mode is 'scaleFill': elif mode is 'scaleFill':
dw, dh = 0.0, 0.0 dw, dh = 0.0, 0.0
new_unpad = (new_shape, new_shape) new_unpad = (new_shape, new_shape)
ratiow, ratioh = new_shape / shape[1], new_shape / shape[0] ratio = new_shape / shape[1], new_shape / shape[0] # width, height ratios
if shape[::-1] != new_unpad: # resize if shape[::-1] != new_unpad: # resize
img = cv2.resize(img, new_unpad, interpolation=interp) # INTER_AREA is better, INTER_LINEAR is faster img = cv2.resize(img, new_unpad, interpolation=interp) # INTER_AREA is better, INTER_LINEAR is faster
top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1)) top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1))
left, right = int(round(dw - 0.1)), int(round(dw + 0.1)) left, right = int(round(dw - 0.1)), int(round(dw + 0.1))
img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=color) # add border img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=color) # add border
return img, ratiow, ratioh, dw, dh return img, ratio, dw, dh
def random_affine(img, targets=(), degrees=10, translate=.1, scale=.1, shear=10): def random_affine(img, targets=(), degrees=10, translate=.1, scale=.1, shear=10):