def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
hand_color1 = np.array(hypes['data']['hand_color1'])
hand_color2 = np.array(hypes['data']['hand_color2'])
hand_color3 = np.array(hypes['data']['hand_color3'])
hand_color4 = np.array(hypes['data']['hand_color4'])
hand_color5 = np.array(hypes['data']['hand_color5'])
background_color = np.array(hypes['data']['background_color'])
gt_hand1 = np.all(gt_image == hand_color1, axis=2) # TODO: may not exist
gt_hand2 = np.all(gt_image == hand_color2, axis=2) # TODO: may not exist
gt_hand3 = np.all(gt_image == hand_color3, axis=2) # TODO: may not exist
gt_hand4 = np.all(gt_image == hand_color4, axis=2) # TODO: may not exist
gt_hand5 = np.all(gt_image == hand_color5, axis=2) # TODO: may not exist
gt_hand = gt_hand1 | gt_hand2 | gt_hand3 | gt_hand4 | gt_hand5
gt_bg = np.all(gt_image == background_color, axis=2) # TODO: may not exist
valid_gt = gt_hand + gt_bg
FN, FP, posNum, negNum = seg.evalExp(gt_hand,
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
FN, FP = np.zeros(thresh.shape), np.zeros(thresh.shape)
posNum, negNum = 0, 0
colors = []
for key in hypes['colors']:
colors.append(np.array(hypes['colors'][key]))
valid_gt = np.all(gt_image == colors[0], axis=2)
for i in range(1, len(colors)):
valid_gt = valid_gt + np.all(gt_image == colors[i], axis=2)
for i in range(len(colors)):
N, P, pos, neg = seg.evalExp(np.all(gt_image == colors[i], axis=2),
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
FN = np.add(FN, N)
FP = np.add(FP, P)
posNum += pos
negNum += neg
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256))/255.0
valid_gt = utils.load_segmentation_mask(hypes, gt_image)[:, :, :-1]
return seg.evalExp(hypes, valid_gt, cnn_image)
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
road_gt = gt_image[:, :, 2] > 0
valid_gt = gt_image[:, :, 0] > 0
FN, FP, posNum, negNum = seg.evalExp(road_gt,
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256))/255.0
road_color = np.array(hypes['data']['road_color'])
background_color = np.array(hypes['data']['background_color'])
gt_road = np.all(gt_image == road_color, axis=2)
gt_bg = np.all(gt_image == background_color, axis=2)
valid_gt = gt_road + gt_bg
FN, FP, posNum, negNum = seg.evalExp(gt_road, cnn_image,
thresh, validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
road_color = np.array(hypes['data']['road_color'])
background_color = np.array(hypes['data']['background_color'])
gt_road = np.all(gt_image == road_color, axis=2)
gt_bg = np.all(gt_image == background_color, axis=2)
valid_gt = gt_road + gt_bg
FN, FP, posNum, negNum = seg.evalExp(gt_road,
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
gt_bg = gt_image[:, :, 0]
gt_hand = gt_image[:, :, 1]
gt_local = gt_image[:, :, 2]
valid_gt = gt_bg | gt_hand
FN, FP, posNum, negNum = seg.evalExp(gt_local,
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
heatmap_color = np.array(hypes['data']['heatmap_color'])
background_color = np.array(hypes['data']['background_color'])
# calculate the probability of an object of interest location in each pixel
gt_obj = np.any(gt_image != background_color, axis=2)
gt_bg = np.all(gt_image == background_color, axis=2)
valid_gt = gt_obj + gt_bg
FN, FP, posNum, negNum = seg.evalExp(gt_obj,
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
def eval_image(hypes, gt_image, cnn_image):
"""."""
thresh = np.array(range(0, 256)) / 255.0
#road_color = np.array(hypes['data']['road_color'])
#background_color = np.array(hypes['data']['background_color'])
gts = []
for color_name, color in hypes['data']['colors'].items():
gts.append(np.all(gt_image == color, axis=2))
valid_gt = np.sum(gts)
# right now this only calculates for the first thing
FN, FP, posNum, negNum = seg.evalExp(gts[0],
cnn_image,
thresh,
validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
import tensorvision
import tensorvision.utils as utils
#evaluate the image
def eval_image(hypes, gt_image, cnn_image):
thresh = np.array(range(0, 256))/255.0
#get the road color for segmentation red, blue/ green
road_color = np.array(hypes['data']['road_color'])
#background_color
background_color = np.array(hypes['data']['background_color'])
gt_road = np.all(gt_image == road_color, axis=2)
gt_bg = np.all(gt_image == background_color, axis=2)
valid_gt = gt_road + gt_bg
FN, FP, posNum, negNum = seg.evalExp(gt_road, cnn_image,
thresh, validMap=None,
validArea=valid_gt)
return FN, FP, posNum, negNum
#resize the label image
def resize_label_image(image, gt_image, image_height, image_width):
#image resize
image = scp.misc.imresize(image, size=(image_height, image_width),
interp='cubic')
shape = gt_image.shape
#label resize
gt_image = scp.misc.imresize(gt_image, size=(image_height, image_width),
interp='nearest')
return image, gt_image
