Exemplo n.º 1
0
class Captioner:
    def __init__(self):
        self.caption_bot = CaptionBot()
        self.prev_caption = ''


    def caption_img(self, frame, get_prev=False):
        if not get_prev:
            img_name = 'to_caption.jpg'
            cv2.imwrite(img_name, frame)
            caption = self.caption_bot.file_caption(img_name)
            self.prev_caption = caption
        return self.prev_caption
Exemplo n.º 2
0
def describe():
    msgtype = request.form['type']
    ##question=request.form['question']
    print(str(msgtype) + "is message ")
    imagefile = request.files['imagefile']
    filename = secure_filename(imagefile.filename)
    imagefile.save(os.path.join(app.config['UPLOAD_FOLDER'], filename))
    print(filename)
    filepath = os.path.join(app.config['UPLOAD_FOLDER'], filename)
    print(filepath)
    if (msgtype == '1'):
        c = CaptionBot()
        print("caption bot running")
        result = c.file_caption(filepath)
        print(result)
        return jsonify({
            'message': result,
            'type': msgtype,
            'filename': filename,
            'filepath': filepath
        })
    elif (msgtype == '2'):
        question = request.form['question']
        ##establish connection
        ##if (question is None) return jsonify({'message':'error , didnt include question','type':'2'})
        response = requests.get(url, headers=header)

        ##upload image
        url1 = 'http://vqa.daylen.com/api/upload_image'
        files = {'file': open(filepath, 'rb')}
        response = requests.post(url1, files=files, headers=header)
        data = json.loads(response.text)
        im_id = data['img_id']
        print im_id
        ## upload question
        url2 = 'http://vqa.daylen.com/api/upload_question'
        #'fa10956c0743670ec5c924ce1523ab16'
        datas = {'img_id': im_id, 'question': question}
        response = requests.post(url2, data=datas, headers=header)
        ##get response
        print(response.text)
        data = json.loads(response.text)
        return jsonify({'message': data[u'answer'], 'type': '2'})
    else:
        return jsonify({'message': 'error in type ', 'type': '1'})
Exemplo n.º 3
0
def captioning(video_id, scenes):
    print('Captioning {}'.format(video_id))
    start_time = datetime.now()
    c = CaptionBot()
    result = []
    cached_path = 'cached/{}.data'.format(video_id)
    if os.path.isfile(cached_path):
        with open(cached_path, 'rb') as cached_file:
            result = pickle.load(cached_file)
        for x in result:
            print('At {}: {}'.format(int(x['start_time']), x['caption']))
    else:
        for x in scenes:
            caption = c.file_caption(x[0])
            print('At {}: {}'.format(int(x[1]), caption))
            result.append({'start_time': x[1], 'caption': caption})
        with open(cached_path, 'wb') as cached_file:
            pickle.dump(result, cached_file)
    print('Captioning: {} seconds'.format(datetime.now() - start_time))
    return result
Exemplo n.º 4
0
from captionbot import CaptionBot

c = CaptionBot()
#msg = c.url_caption("https://beebom-redkapmedia.netdna-ssl.com/wp-content/uploads/2016/01/Reverse-Image-Search-Engines-Apps-And-Its-Uses-2016.jpg")
#msg2 = c.url_caption("https://upload.wikimedia.org/wikipedia/commons/thumb/6/63/Collection_of_military_aircraft.jpg/1200px-Collection_of_military_aircraft.jpg")
msg2 = c.file_caption("/Users/khanh/Desktop/bbb.jpg")
print(msg2)
Exemplo n.º 5
0
import cv2
from captionbot import CaptionBot

c = CaptionBot()
cap = cv2.VideoCapture(0)
while True:
    r, f = cap.read()
    if r == True:
        cv2.imshow("Stream", f)
        key = cv2.waitKey(1)
        if key == ord('a'):
            print("Generating Caption...")
            cv2.imwrite('image.jpg', f)
            caption = c.file_caption('/home/aditya/Hack-a-bit2019/' +
                                     'image.jpg')
            print(caption)
        elif key == 27:
            break
    else:
        continue
cv2.destroyAllWindows()
cap.release()
json_files = 0

for f in os.listdir('data/frames/'):
    parsed = f.split(".")
    if parsed[1] == 'png':
        frame_files += 1
    else:
        json_files += 1

index = json_files + 1

print "Total frames:", str(frame_files)
print "Total JSONs:", str(json_files)
print "To process now:", str(frame_files - json_files)
print "Starting at frame:", str(index)

for i in range(index, frame_files):
    #c.url_caption('your image url here')
    data = c.file_caption('data/frames/' + str(i) + '.png')
    print "Frame " + str(i) + " :: " + data

    json_object['data']['caption'] = data
    json_object['data']['filename'] = 'data/frames/' + str(i) + '.png'
    json_object['data']['film-frame'] = i

    os.system('touch data/frames/' + str(i) + ".json")

    with open("data/frames/" + str(i) + ".json", 'r+') as json_data:
        json.dump(json_object, json_data, indent=2)

print "Finished!"