README

Common library with python code snippets used frequently.

Libraries Implemented

• OpenCV
• Python multiprocessing
• VideoManager
• Camera capture utils
• Motion detector
• common Inference clients
• Usb camera mount path finder

Installation

pip install --user git+https://gitlab.com/_macherlabs/visonlibs/common.git

Video Recorder annotation usage example

video_writer = VideoWriter(filename='pc-rgb.'+'.avi',
                            fourcc = 'XVID',
                            fps = 1,
                            frameSize = (640,480),
                            base_dir = '/LFS/depth_vids',
                            max_sizeMB = 2,
                            cleanupDays = 3,
                            annotations= True,
                            annotations_header=['frame_number','xmin','ymin','xmax','ymax'])
 img=np.ones((640,480)).astype('uint8')
 annotation = ["10","20","30","40"]
 video_writer.write(img,annotation)
 

USB cam_finder usage

Returns the mount path for usb camera say ('/dev/video0')

cam_mount_path=find_usb_cam_path(
                      VENDOR_ID='1234',
                      MODEL_ID ='12',
                      SERIAL_ID='123')

cam utils usage

Wrapper for reading camera streams
supported streams-rtsp,http,usb

    # Camera config
    cam_config={
            "name":"testcamera",
            "url":"rtsp://test.avi",
            "_id":"1234",
            "params":
                {
                    "readMethod":"video", # if video, will read images from saved video via video manager, else rtsp
                    "motionEnabled":True # if true, will save videos only of frames with motion
                }
        }

    # Video config
    vid_config = 
    {
        "base_dir":"/Users/saurabh_veda/workspace_kumar/new_test"
    }
    
    cap = cap_rtsp(cam_config,vid_config)
    
    while True:
        ret,image=cap.read()
        image_time = cap.get_process_time() # Actual time when image was recorded
        if ret:
            cv2.imshow("vid",image)
            cv2.waitKey(1)
        

Motion Detector usage

Check if motion is observed in camera stream

    config ={
      "threshold_scale_factor":0.1, threshold above which event is registered
      "start_at":"12:02am", # start time for motion detection
      "end_at":"11:59pm", # end time for motion detection
      "roi":{"x":10,"y":10,"x1":20,"y1":50}, # check motion only in roi
      "timeCheck":False # if true checks motion during specific time limit only
    }
    detector=motionDetector(config)
    cap=cv2.VideoCapture(0)
    
    while True:
        ret,img=cap.read()
        if ret:
            det= detector.detect(img)

Inference Detector client usage

Client for detection inferences

    config={"endpoint":"http://localhost",
            "port":9000,
            "route":"detect"}
    detector=detectorClient(config)
    while True:
        img=np.random.rand(300,300,3)
        det= detector.detect(img)
        print("dets",det)

Inference Face recognizer client usage

Client for insight face embedding calculator

    config={"endpoint":"http://localhost",
            "apiPort":9000,
            "mlPort":6500,
            "api_version":"api/v1",
            "service":"recognition",
            "api_key":"00000000-0000-0000-0000-000000000002",
            "maxFailedAttempts":5000,
            "matchThreshold":0.6
            }

    recognizer=recognizerClient(config)
    while True:
        img=np.random.rand(300,300,3)
        embedding= recognizer.get_encodings(img)
        print("dets",det)

Video manager usage

Video manager for writing and reading saved videos

    manager = VideoManager(filename='test.avi',
                                        fourcc = 'XVID', #encoding for saved video
                                        fps = 30, # fps of saved video
                                        frameSize = (640,480), # resolution of saved video
                                        base_dir = '/Users/saurabh_veda/workspace_kumar/test_videos', # directory where videos will be saved
                                        max_sizeMB = 500, # max size after video will be released
                                        max_seconds= 60, # max duration of video after which video will be released
                                        max_period = 60 # max time period after which video will be released
                                        cleanupDays = 1 # no of days video will be retained
                                        )
    
    while True:
        ret,img=manager.read()
        if ret:
            cv2.imshow("vid",img)
            cv2.waitKey(1)