Object tracking basics in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to initialize the OpenCV tracker.

Computer Vision

tracker = cv2.TrackerCSRT_create()  # Create CSRT tracker
success, frame = video.read()
bbox = cv2.selectROI(frame, False)
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, [1])

Drag options to blanks, or click blank then click option'

Avideo

Bframe

Cbbox

Dsuccess

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the whole frame instead of the bounding box to tracker.init()

Using video or success variables instead of bbox

The tracker needs the bounding box (bbox) to know what object to track in the initial frame.

2fill in blank

medium

Complete the code to update the tracker and get the new bounding box.

Computer Vision

success, frame = video.read()
success, [1] = tracker.update(frame)
if success:
    p1 = (int(bbox[0]), int(bbox[1]))
    p2 = (int(bbox[0] + bbox[2]), int(bbox[1] + bbox[3]))
    cv2.rectangle(frame, p1, p2, (255,0,0), 2, 1)

Drag options to blanks, or click blank then click option'

Aframe

Bbbox

Csuccess

Dvideo

Attempts:

3 left

💡 Hint

Common Mistakes

Using frame instead of bbox to store the updated location

Confusing success with bbox in the assignment

The tracker.update() returns success and the new bounding box (bbox) for the tracked object.

3fill in blank

hard

Fix the error in the code to correctly select the ROI for tracking.

Computer Vision

frame = video.read()[1]
bbox = cv2.selectROI([1], False)

Drag options to blanks, or click blank then click option'

Asuccess

Bvideo

Ctracker

Dframe

Attempts:

3 left

💡 Hint

Common Mistakes

Passing the video capture object instead of the frame image

Passing the tracker object instead of the frame

cv2.selectROI requires the current frame image to select the region of interest (ROI).

4fill in blank

hard

Fill both blanks to create a dictionary of trackers and update each one.

Computer Vision

trackers = {}
trackers['car'] = cv2.TrackerKCF_create()
trackers['car'].init(frame, bbox_car)
trackers['person'] = cv2.TrackerMIL_create()
trackers['person'].init(frame, bbox_person)

for name, tracker in trackers.items():
    success, [1] = tracker.update(frame)
    if success:
        p1 = (int([2][0]), int([2][1]))

Drag options to blanks, or click blank then click option'

Abbox

Bbbox_car

Cbbox_person

Dframe

Attempts:

3 left

💡 Hint

Common Mistakes

Using bbox_car or bbox_person inside the loop instead of the updated bbox

Using frame instead of bbox for the bounding box variable

The update returns the new bounding box which should be stored in bbox to draw the rectangle.

5fill in blank

hard

Fill all three blanks to create a dictionary comprehension that tracks objects with bounding boxes larger than 5000 pixels.

Computer Vision

large_objects = {name: bbox for name, bbox in object_bboxes.items() if ([1] * [2]) [3] 5000}

Drag options to blanks, or click blank then click option'

Abbox[2]

Bbbox[3]

C>

D<

Attempts:

3 left

💡 Hint

Common Mistakes

Using x or y instead of width and height for area calculation

Using less than instead of greater than for filtering

The width (bbox[2]) times height (bbox[3]) gives the area. We filter for area greater than 5000.