def obj_detect():
if 'file' not in request.files:
return 'no image file uploaded'
File = request.files['file']
# can check extension here before detecting
if File.filename == '':
return 'no file name provided'
if File and allowed_file(File.filename):
fn = secure_filename(File.filename)
File.save(os.path.join(app.config['UPLOAD_FOLDER'], fn))
imagePath = os.path.join(app.config['UPLOAD_FOLDER'], fn)
detect(imagePath)
resp = send_from_directory(UPLOAD_FOLDER, fn)
resp.headers['filename'] = fn
os.remove(imagePath)
return resp
else:
return 'file type is not supported'
def label():
req_data = request.get_json(force=True)
print(req_data)
root_save_dir = "{}/projects/{}/{}".format(drive, req_data["projectName"],
req_data["dataType"])
save_dir = "{}/images".format(root_save_dir)
Path(save_dir).mkdir(parents=True, exist_ok=True)
imgdata = base64.b64decode(req_data["imageBase64"])
filename = next_path("{}/{}_%s.png".format(save_dir, req_data["label"]))
with open(filename, "wb") as f:
f.write(imgdata)
f.close()
print("saved: {}".format(filename))
save_label(req_data["label"], root_save_dir)
result = []
if req_data["autoMakeSense"]:
result = detect(filename, req_data["objectConfidenceThreshold"])
if len(result) > 0:
lines = ""
for item in result:
for v in item:
lines += str(v) + " "
lines = lines.strip() + "\n"
with open(filename + ".pre", "w") as f:
f.write(lines.strip())
f.close()
return json.dumps([labels, result])
def yolo_detect(net, img):
detections = yolo.detect(net, img)
filtered = yolo.filter(img, detections)
detected, img = yolo.apply_nms(img, filtered, 0.4, 0.3)
if detected:
print("Meteoro detectado!\nImagem salva em detections/")
timestmp = str(time.time())
name = "detections/" + timestmp.split(".")[0] + timestmp.split(".")[1]
cv2.imwrite(name, img)
return detected
def exportFrame():
event = controller.step(dict(action='Initialize', continuous=True))
cv2.imwrite('frame.jpg', event.cv2img)
yolo.detect()
import yolo yolo.detect()
import os
import cv2
import yolo
image = yolo.detect('/yolo/test.png')
cv2.imwrite('result.png', image)
[1024, 490], [1024, 544]], np.int32)
pts = pts.reshape((-1, 1, 2))
imgL_detect = imgL.copy()
cv2.fillConvexPoly(imgL_detect, pts, (0, 0, 0))
# Cover up car bonnet in display version of disparity map
pts2 = np.array([[-55, 544], [-55, 515], [385, 388], [630, 388],
[1024, 475], [1024, 544]], np.int32)
pts2 = pts2.reshape((-1, 1, 2))
cv2.fillConvexPoly(disparity_display, pts2, (0, 0, 0))
windowName = 'left image'
cv2.namedWindow(windowName, cv2.WINDOW_NORMAL)
cv2.resizeWindow(windowName, imgL.shape[1], imgL.shape[0])
# Perform YOLO detection
classIDs, confidences, boxes = yolo.detect(imgL_detect,
allowed_classes)
# Sort objects according to disparity
for i in range(0, len(boxes)):
box = boxes[i]
left = box[0]
top = box[1]
width = box[2]
height = box[3]
roi = disparity[max(top, 0):top + height,
max(left, 0):left + width]
disp = np.mean(roi[roi >= 0])
box.append(disp)
box.append(classIDs[i])
boxes = sorted(boxes, key=lambda x: x[4], reverse=True)
min_depth = 0
def main():
while True:
# imposta il tempo di attesa fra un conteggio di persone e l'altro
print("Orario di inizio detection: " + str(datetime.now()))
t_end = datetime.now().replace(microsecond=0) + timedelta(
seconds=DETECTIONS_INTERVAL)
print("Orario di fine detection: " + str(t_end))
# leggi il file json contenente i dati delle webcam
with open(PATH_WEBCAM_JSON) as f:
json_data = json.load(f)
# scorri fra tutte le webcams presenti nel file json
for webcam in json_data["webcams"]:
print("Raccolta dati per: " + webcam["location"] + ", " +
webcam["city"])
# ottieni la temperatura e il meteo al tempo dell'acquisizione;
try:
temperature, weather_description = get_current_weather(
webcam["latitude"], webcam["longitude"])
except:
print('Exception nel weather')
logging.info('Eccezione ore: ' + str(datetime.now()))
logging.error(sys.exc_info())
continue
#imposta l'orario e la data di acquisizione
current_time = current_date = datetime.now()
try:
video_link = fetch_read_m3u8(webcam["link"],
webcam["url_prefix"])
except:
print("Failed to fetch/read m3u8")
# continue
# scarica il video dal link appena ricavato
try:
#urllib.request.urlretrieve(video_link, PATH_VIDEOS+"Video" + ".ts")
frame_is_read, frame = extract_frame_from_video_url(video_link)
print(frame_is_read)
except:
print('Exception: video non disponibile: ' +
webcam["location"])
logging.info('Eccezione ore: ' + str(datetime.now()))
logging.error(sys.exc_info())
continue
try:
frame_part = cut_frame_in_six(frame)
except:
print('Exception')
logging.info('Eccezione ore: ' + str(datetime.now()))
logging.error(sys.exc_info())
continue
persone_contate = 0
# conta le persone in ogni sottoframe
for frame in frame_part:
persone_contate = persone_contate + detect(frame)
print("Persone: " + str(persone_contate))
data = {
'id_webcam': webcam["id_webcam"],
'city': webcam["city"],
'location': webcam["location"],
'latitude': webcam["latitude"],
'longitude': webcam["longitude"],
'numPeople': persone_contate,
'date': current_date.strftime('%Y-%m-%d %H:%M:%S.%f'),
'time': current_time.strftime('%Y-%m-%d %H:%M:%S.%f'),
'type': 0,
'weather_description': weather_description,
'temperature': temperature,
'day_of_week': datetime.now().date().weekday()
}
producer.send('gdp', value=data)
print("ho fatto kafka")
# inserisci i risultati nel db
print("waiting for: " + str((t_end - datetime.now()).total_seconds()))
time.sleep((t_end - datetime.now()).total_seconds())
def laptop_recommandations(product):
c = [categories[product].split(',')[0:4]]
for i in range(len(product_names)):
if categories[product].split(',')[0:3] == categories[i].split(
',')[0:3] and categories[i].split(',')[0:4] not in c:
c.append(categories[i].split(',')[0:4])
del c[4]
del c[4]
del c[4]
del c[-1]
rec = []
distances = []
for i in range(len(product_names)):
if product_names[i] != product_names[product] and str(
product_descriptions[i]) != 'nan' and str(
dsc_image_urls[i]) != 'nan' and categories[i].split(
',')[0:4] == c[0]:
d = 1 - similarity_test.evaluateSimilarity(product, i, data)
distances.append((d, i))
distances = sorted(distances)
rec = [id[distances[0][1]]]
titles = [product_names[distances[0][1]]]
n = 5 #number of similar products
i = 1
while len(titles) < n:
if product_names[distances[i][1]] not in titles:
titles.append(product_names[distances[i][1]])
rec[0] += ' ' + id[distances[i][1]]
i += 1
rec.append('')
for k in range(1, len(c)):
distances = []
for i in range(len(product_names)):
if product_names[i] != product_names[product] and str(
product_descriptions[i]) != 'nan' and str(
dsc_image_urls[i]) != 'nan' and categories[i].split(
',')[0:4] == c[k]:
d = 1 - similarity_test.evaluateSimilarity(product, i, data)
distances.append((d, i))
distances = sorted(distances)
admissibles = []
j = 0
for l in range(5):
img = requests.get(dsc_image_urls[distances[j][1]]).content
name = 'product.jpg'
with open(yolo.image_path + name, 'wb') as handler:
handler.write(img)
y = yolo.detect(name)
while product_names[distances[j][1]] in titles or y == [
'laptop'
] or 'tvmonitor' in y:
j += 1
img = requests.get(dsc_image_urls[distances[j][1]]).content
name = 'product.jpg'
with open(yolo.image_path + name, 'wb') as handler:
handler.write(img)
y = yolo.detect(name)
admissibles.append(distances[j][1])
j += 1
i1, i2 = random.sample(admissibles, 2)
rec[-1] += ' ' + id[i1] + ' ' + id[i2]
titles.append(product_names[i1])
titles.append(product_names[i2])
rec[-1] = rec[-1][1:]
return rec
quotechar='"',
escapechar=' ',
quoting=csv.QUOTE_NONE)
for i in range(len(product_names)):
if categories[
i] == 'Root Category, Sporting Goods, Outdoor Recreation, Cycling, Bikes' and str(
dsc_image_urls[i]) != 'nan' and str(
product_descriptions[i]) != 'nan':
img = requests.get(dsc_image_urls[i]).content
img_name = product_names[i].replace('"', '').replace(
"/", "").replace("\\", "").replace(":", "").replace(
"*", "").replace("?", "").replace("<", "").replace(
">", "").replace("|", "").replace(".", "") + '.jpg'
with open(yolo.image_path + img_name, 'wb') as handler:
handler.write(img)
if 'bicycle' in yolo.detect(img_name):
rec = bike_recommandations(i, img_name)
if len(rec[1].split(' ')) == 6:
print(i)
r.writerow([id[i]] + rec)
with open(r'ps4.csv', mode='w', newline="") as csvfile:
r = csv.writer(csvfile,
delimiter=',',
quotechar='"',
escapechar=' ',
quoting=csv.QUOTE_NONE)
for i in range(len(product_names)):
if categories[
i] == 'Root Category, Gaming, Playstation, PlayStation 4, Consoles' and str(
dsc_image_urls[i]) != 'nan' and str(
