def test():
image = mpimg.imread('./test_images/test1.jpg')
draw_image = np.copy(image)
image = image.astype(np.float32)*255
windows = slide_window(image, x_start_stop=[None, None], y_start_stop=[350, None],
xy_window=(96, 96), xy_overlap=(0.5, 0.5))
hot_windows = search_windows(image, windows, svc, X_scaler, color_space=color_space,
spatial_size=spatial_size, hist_bins=hist_bins,
orient=orient, pix_per_cell=pix_per_cell,
cell_per_block=cell_per_block,
hog_channel=hog_channel, spatial_feat=spatial_feat,
hist_feat=hist_feat, hog_feat=hog_feat)
window_img = draw_boxes(draw_image, hot_windows, color=(0, 0, 255), thick=6)
plt.imshow(window_img)
def test_image():
img = mpimg.imread('./test_images/test6.jpg')
rectangles = []
scale = 1.0
y_start_stop = [350, 700]
x_start_stop = [None, None]
rectangles.append(find_cars(img, x_start_stop, y_start_stop, scale ,color_space, svc,
X_scaler, orient, pix_per_cell, cell_per_block, spatial_size, hist_bins ,show_all_rectangles=False))
scale = 1.5
y_start_stop = [400, None]
x_start_stop = [None, None]
rectangles.append(find_cars(img, x_start_stop, y_start_stop, scale ,color_space, svc,
X_scaler, orient, pix_per_cell, cell_per_block, spatial_size, hist_bins,show_all_rectangles=False))
scale = 2.0
y_start_stop = [450, None]
x_start_stop = [None, None]
rectangles.append(find_cars(img, x_start_stop, y_start_stop, scale ,color_space, svc,
X_scaler, orient, pix_per_cell, cell_per_block, spatial_size, hist_bins,show_all_rectangles=False))
scale = 2.5
y_start_stop = [500, None]
x_start_stop = [None, None]
rectangles.append(find_cars(img, x_start_stop, y_start_stop, scale ,color_space, svc,
X_scaler, orient, pix_per_cell, cell_per_block, spatial_size, hist_bins,show_all_rectangles=False))
# apparently this is the best way to flatten a list of lists
rectangles = [item for sublist in rectangles for item in sublist]
test_img_rects = draw_boxes(img, rectangles, color=(0, 0, 255), thick=2)
plt.figure(figsize=(10,10))
plt.imshow(test_img_rects)
heatmap_img = np.zeros_like(img[:,:,0])
heatmap_img = add_heat(heatmap_img, rectangles)
heatmap_img = apply_threshold(heatmap_img, 1)
plt.figure(figsize=(10,10))
plt.imshow(heatmap_img, cmap='hot')
labels = label(heatmap_img)
plt.figure(figsize=(10,10))
plt.imshow(labels[0], cmap='gray')
draw_img = draw_labeled_bboxes(img, labels)
plt.figure(figsize=(10,10))
plt.imshow(draw_img)
def decodeFrame(image, image_w, image_h, model, yhat, imageRead):
boxes = list()
for i in range(len(yhat)):
boxes += dF.decode_netout(yhat[i][0], anchors[i], propability, input_h, input_w)
dF.correct_yolo_boxes(boxes, image_h, image_w, input_h, input_w)
dF.do_nms(boxes, 0.5)
v_boxes, v_labels, v_scores = f.get_boxes(boxes, labels, propability)
imageFinal = f.draw_boxes(imageRead, v_boxes, v_labels, v_scores, 0.5)
return imageFinal
def decodeFrame(image, image_w, image_h, model, yhat, imageFile):
boxes = list()
for i in range(len(yhat)):
# Decodeer de bounding boxes (de rechthoekjes om het herkende heen)
boxes += dF.decode_netout(yhat[i][0], anchors[i], propability, input_h,
input_w)
# Zet de shape en size van de boxes weer om naar het originele plaatje ipv het 416x416 plaatje
dF.correct_yolo_boxes(boxes, image_h, image_w, input_h, input_w)
dF.do_nms(
boxes, 0.5
) # zorg ervoor dat overlappende bounding boxes weg gehaald worden( 0.5 staat voor bounding boxes die 50% over een komen)
v_boxes, v_labels, v_scores = f.get_boxes(boxes, labels, propability)
f.br()
print("Die tensorflow error net:")
print(
"'WARNING:tensorflow:No training configuration found in the save file, so the model was *not* compiled. Compile it manually.'"
)
print("Is niks ergs, maak geen zorgen")
f.br()
# eventjes de aparte boxen printen
for i in range(len(v_boxes)):
print(v_labels[i], v_scores[i])
imageFinal = f.draw_boxes(imageFile, v_boxes, v_labels, v_scores, 0.5)
print(imageFinal)
cv2.imshow("window", imageFinal)
cv2.waitKey(0)
cv2.destroyAllWindows()
cv2.imwrite(imageFile[:-4] + '_detected' + imageFile[-4:],
(imageFinal).astype('uint8'))
cars = list()
fps = 1
try:
while True:
beginning = time.time() #For FPS calculations
_, frame = video_capture.read()
if frame is None:
sess.close()
exit("End of video file!")
frame2, boxes, scores = detect_objects(frame, sess, detection_graph)
#boxes : [upper-y, upper-x, lower-y, lower-x] : float percentage of image resolution
functions.track_cars(boxes, scores, cars, width, height, fps)
if args.display == "y":
functions.draw_boxes(cars, frame2)
cv2.imshow('image', frame2)
cv2.waitKey(1)
num_cars = functions.persistence(cars, width, height)
fps = 1 / (time.time() - beginning)
print("FPS:", fps)
except KeyboardInterrupt:
print('interrupted!')
windows,
svc,
X_scaler,
color_space=color_space,
spatial_size=spatial_size,
hist_bins=hist_bins,
orient=orient,
pix_per_cell=pix_per_cell,
cell_per_block=cell_per_block,
hog_channel=hog_channel,
spatial_feat=spatial_feat,
hist_feat=hist_feat,
hog_feat=hog_feat)
window_img = functions.draw_boxes(draw_img,
hot_windows,
color=(0, 0, 255),
thick=6)
images.append(window_img)
titles.append('')
print(round(time.time() - t1, 2), "seconds to process one image searching",
len(windows), "windows.")
fig = plt.figure(figsize=(17, 18)) #, dpi=300)
functions.visualize(fig, 3, 2, images, titles)
# CALCULATE HOG FEATURES ONLY ONCE
out_images = []
out_maps = []
out_titles = []
out_boxes = []
boxes = Predictor.boxes.get()
edgearray = Predictor.boxes.get()
video_frame = Predictor.boxes.get()
if boxes is None:
exit()
#with open("fps.log", "a") as myfile: #Write fps to file for logging
# myfile.write(str(args.gen_threads)+
# " "+str(args.pred_threads)+
# " "+str(256/(time.time()-bbeginning))+
# "\n")
#exit(256/(time.time()-bbeginning))
fps = 1 / (time.time() - beginning)
print("FPS: ", int(fps))
network.sendBoxes(video_frame, boxes)
if args.visualize == 'y':
frame = draw_boxes(boxes, edgearray)
cv2.imshow('image', frame)
cv2.imshow('edgemap', edgearray)
cv2.imshow('original', video_frame)
cv2.waitKey(10)
frames = frames + 1
total_fps = total_fps + fps
except KeyboardInterrupt:
exit()
fig4.savefig('./examples/car_notcar_hog.jpg')
image = mpimg.imread('./test_images/test1.jpg')
window_size = 64
search_boxes = [(1, 0, 199, 400, 530), (1.5, 200, 399, 400, 600),
(2, 400, 599, 400, 700), (2.5, 600, 799, 400, 700),
(3, 800, 999, 400, 700)]
for scale, xstart, xstop, ystart, ystop in search_boxes:
windows = slide_window(image,
x_start_stop=[xstart, xstop],
y_start_stop=[ystart, ystop],
xy_window=(int(scale * window_size),
int(scale * window_size)),
xy_overlap=(0, 0))
image = draw_boxes(image, windows, color=(0, 0, 255), thick=6)
plt.imsave('./examples/sliding_windows.jpg', image)
search_boxes = [(1, 400, 530), (1.5, 400, 600), (2, 400, 700), (2.5, 400, 700),
(3, 400, 700)]
test_images = glob.glob('./test_images/test*.jpg')
fig1 = plt.figure(figsize=(8, 12))
fig2 = plt.figure(figsize=(8, 18))
fig3 = plt.figure(figsize=(8, 12))
for idx, file in enumerate(test_images):
img = mpimg.imread(file)
total_heat = np.zeros_like(img[:, :, 0]).astype(np.float)
for scale, ystart, ystop in search_boxes:
bbox_list = find_cars(img,
colorspace,
hist_bins=for_pickle['hist_bins']
orient=for_pickle['orient']
pix_per_cell=for_pickle['pix_per_cell']
cell_per_block=for_pickle['cell_per_block']
hog_channel=for_pickle['hog_channel']
spatial_feat=for_pickle['spatial_feat']
hist_feat=for_pickle['hist_feat']
hog_feat=for_pickle['hog_feat']
show_scalar=True
#test_img_bgr = cv2.imread('./test_images/test1.jpg')
#
#test_img = convert_color(test_img_bgr,color_space='RGB')
ffc,axfc=plt.subplots(1,2)
test_img = mpimg.imread('../test_images/test1.jpg')
print("img shape",test_img.shape)
ystart = 400
ystop = 656
scale = 1.5
print("spatial enabled:",spatial_feat,"Hist enabled:",hist_feat)
bbox = find_cars(test_img, ystart, ystop, scale, color_space, hog_channel, svc,
X_scaler, orient, pix_per_cell, cell_per_block, spatial_size, hist_bins,spatial_feat,hist_feat,axfc=axfc,show_scalar=show_scalar)
print(len(bbox), 'bboxes found in image')
show_scalar=False
im_bbox=draw_boxes(test_img, bbox, color=(0, 0, 255), thick=6)
f,p=plt.subplots()
p.imshow(im_bbox)
# combine all heatmaps
heatmap = heatmap_far + heatmap_middle + heatmap_near
# apply threshold to the heatmap
heatmap = functions.apply_threshold(heatmap, threshold)
# apply labels to the heatmap
labels = label(heatmap)
# draw labels on an image copy
labeled_image = functions.draw_labeled_bboxes(draw_img, labels)
# Sum up all windows that were searched
searched_windows = searched_windows_far + searched_windows_middle + searched_windows_near
# Draw boxes in the image around all hot windows
window_img = functions.draw_boxes(draw_img, windows, color=(0, 255, 0), thick=6)
# Append titles for all 4 version of the image
titles.append(img_src[-12:])
titles.append(img_src[-12:])
titles.append(img_src[-12:])
titles.append(img_src[-12:])
# Append Images with windows to image list
images.append(window_img)
# Append Images with windows to image list
images.append(heatmap)
# Append Images with labels to image list
images.append(labels[0])
# Append Images with labeled boxes to image list
images.append(labeled_image)