EfficientNet scaling in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the EfficientNet model from TensorFlow Keras applications.

Computer Vision

from tensorflow.keras.applications import [1]

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

AEfficientNetB0

BResNet50

CVGG16

DMobileNetV2

Attempts:

3 left

💡 Hint

Common Mistakes

Importing a different model like ResNet50 or VGG16 instead of EfficientNetB0.

The EfficientNetB0 model is imported from tensorflow.keras.applications to use EfficientNet architecture.

2fill in blank

medium

Complete the code to create an EfficientNetB0 model with imagenet weights and include the top classification layer.

Computer Vision

model = EfficientNetB0(weights=[1], include_top=True)

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

ANone

B'imagenet'

C'random'

D'custom'

Attempts:

3 left

💡 Hint

Common Mistakes

Using None or other strings which do not load pretrained weights.

Using weights='imagenet' loads pretrained weights from ImageNet dataset for EfficientNetB0.

3fill in blank

hard

Fix the error in the code to scale EfficientNet width and depth correctly using compound scaling.

Computer Vision

def scale_effnet(width_coefficient, depth_coefficient, [1]):

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

Aresolution

Bdropout_rate

Cbatch_size

Dinput_shape

Attempts:

3 left

💡 Hint

Common Mistakes

Using dropout_rate or batch_size which are unrelated to scaling input size.

The resolution parameter is needed to scale the input image size in EfficientNet compound scaling.

4fill in blank

hard

Fill both blanks to complete the function that calculates scaled filters and blocks for EfficientNet.

Computer Vision

def scale_filters(filters, [1]):
    return int(filters * [2])

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

Awidth_coefficient

Bdepth_coefficient

Dresolution

Attempts:

3 left

💡 Hint

Common Mistakes

Using depth_coefficient or resolution to scale filters.

The width_coefficient is used to scale the number of filters in EfficientNet.

5fill in blank

hard

Fill all three blanks to complete the function that scales the number of blocks in EfficientNet using depth coefficient and rounds up.

Computer Vision

import math

def scale_blocks([1], [2]):
    scaled = blocks * depth_coefficient
    return math.ceil(scaled)

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

Ablocks

Bdepth_coefficient

Cmath

Dwidth_coefficient

Attempts:

3 left

💡 Hint

Common Mistakes

Using width_coefficient instead of depth_coefficient.

Not importing math module.

The function takes blocks and depth_coefficient to scale and round up the number of blocks.