Object detection from aerial view in Drone Programming - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load an aerial image for object detection.

Drone Programming

image = cv2.imread([1])

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

Aimage.jpg

Bimg.png

Cload_image()

D'aerial_view.jpg'

Attempts:

3 left

💡 Hint

Common Mistakes

Forgetting quotes around the filename.

Using a function call instead of a filename string.

The correct way to load the aerial image is by specifying the filename as a string inside quotes.

2fill in blank

medium

Complete the code to convert the aerial image to grayscale for processing.

Drone Programming

gray_image = cv2.cvtColor(image, [1])

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

Acv2.COLOR_BGR2GRAY

Bcv2.COLOR_BGR2RGB

Ccv2.COLOR_RGB2GRAY

Dcv2.COLOR_GRAY2BGR

Attempts:

3 left

💡 Hint

Common Mistakes

Using COLOR_BGR2RGB instead of grayscale conversion.

Using COLOR_GRAY2BGR which is the reverse conversion.

To convert a BGR image to grayscale, use cv2.COLOR_BGR2GRAY.

3fill in blank

hard

Fix the error in the code to detect objects using a pre-trained model.

Drone Programming

detections = model.[1](gray_image)

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

Atransform

Bdetect

Cfit

Dpredict

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'predict' which is common in classification but not detection.

Using 'fit' which is for training, not detection.

The correct method to detect objects is 'detect', not 'predict' or others.

4fill in blank

hard

Fill both blanks to filter detected objects by confidence and draw bounding boxes.

Drone Programming

for obj in detections:
    if obj.confidence [1] 0.5:
        cv2.rectangle(image, obj.bbox[[2]], obj.bbox[2], (0,255,0), 2)

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

A>

B<

C0

D1

Attempts:

3 left

💡 Hint

Common Mistakes

Using less than instead of greater than for confidence.

Using wrong bounding box index causing errors.

We filter objects with confidence greater than 0.5 and use index 0 for the bounding box start point.

5fill in blank

hard

Fill all three blanks to create a dictionary of detected object labels and their counts.

Drone Programming

counts = {}
for obj in detections:
    label = obj.[1]
    counts[label] = counts.get(label, [2]) + [3]

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

Alabel

B0

C1

Dconfidence

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'confidence' instead of 'label' for dictionary keys.

Starting counts at 1 instead of 0 causing wrong counts.

We use obj.label to get the label, default count 0, and add 1 for each detected object.

Practice

1. What is the main goal of object detection from an aerial view using drones?

easy

A. To stream live video to a mobile device

B. To control the drone's flight path automatically

C. To improve battery life of the drone

D. To find and locate objects in images taken from above

Solution

1. Step 1: Understand object detection purpose

   Object detection means finding and locating objects in images or videos.

2. Step 2: Apply to aerial images

   From aerial view, drones capture images from above to detect objects like cars or trees.

3. Final Answer:

   To find and locate objects in images taken from above -> Option D

4. Quick Check:

   Object detection = find objects [OK]

Hint: Object detection means finding objects in images [OK]

Common Mistakes:

• Confusing object detection with drone control
• Thinking it improves battery life
• Mixing detection with video streaming

2. Which Python library is commonly used to build object detection models for aerial images?

easy

A. NumPy

B. Matplotlib

C. OpenCV

D. Pandas

Solution

1. Step 1: Identify libraries for image processing

   OpenCV is widely used for image and video processing, including object detection.

2. Step 2: Compare with other options

   NumPy handles arrays, Matplotlib plots data, Pandas manages tables, but OpenCV is best for images.

3. Final Answer:

   OpenCV -> Option C

4. Quick Check:

   Image processing = OpenCV [OK]

Hint: OpenCV is for image tasks like detection [OK]

Common Mistakes:

• Choosing NumPy for detection instead of OpenCV
• Confusing plotting library Matplotlib with detection
• Selecting Pandas which is for data tables

3. Given this Python snippet for detecting objects in an aerial image, what will be printed?

import cv2
image = cv2.imread('aerial.jpg')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
faces = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
detected = faces.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5)
print(len(detected))

medium

A. Always prints 0 because aerial images have no faces

B. Number of detected faces in the image

C. Error because 'haarcascade_frontalface_default.xml' is missing

D. Prints the image size

Solution

1. Step 1: Understand code function

   The code loads an image, converts it to grayscale, and uses a face detector to find faces.

2. Step 2: Interpret print output

   It prints the number of detected faces found by detectMultiScale.

3. Final Answer:

   Number of detected faces in the image -> Option B

4. Quick Check:

   detectMultiScale returns detected objects count [OK]

Hint: detectMultiScale returns detected object count [OK]

Common Mistakes:

• Assuming error if XML file is present
• Thinking aerial images never have faces
• Confusing output with image size

4. This code tries to detect cars from drone images but fails. What is the main error?

import cv2
car_cascade = cv2.CascadeClassifier('cars.xml')
image = cv2.imread('drone_view.jpg')
cars = car_cascade.detectMultiScale(image, 1.1, 3)
print(len(cars))

medium

A. detectMultiScale needs a grayscale image, not color

B. CascadeClassifier file 'cars.xml' is not loaded correctly

C. cv2.imread should read video, not image

D. print(len(cars)) should be print(cars)

Solution

1. Step 1: Check input type for detectMultiScale

   detectMultiScale requires a grayscale image, but code uses color image.

2. Step 2: Fix image conversion

   Convert image to grayscale using cv2.cvtColor before detection.

3. Final Answer:

   detectMultiScale needs a grayscale image, not color -> Option A

4. Quick Check:

   detectMultiScale input = grayscale image [OK]

Hint: detectMultiScale needs grayscale input [OK]

Common Mistakes:

• Ignoring image color format
• Assuming XML file loading error without checking
• Confusing image and video input

5. You want to improve drone object detection accuracy for small objects far away. Which approach helps most?

hard

A. Use a higher resolution camera and train a deep learning model

B. Reduce the drone's flying height to get closer images

C. Increase the drone's speed to cover more area quickly

D. Use simpler features like color histograms only

Solution

1. Step 1: Understand problem with small distant objects

   Small objects are hard to detect; higher resolution images capture more detail.

2. Step 2: Choose best method to improve detection

   Deep learning models learn complex features and work well with high-res images, improving accuracy.

3. Final Answer:

   Use a higher resolution camera and train a deep learning model -> Option A

4. Quick Check:

   High-res + deep learning = better detection [OK]

Hint: Better images + deep learning improve detection [OK]

Common Mistakes:

• Thinking flying lower is always best
• Believing faster speed improves accuracy
• Using only simple features limits detection