3D object detection in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a 3D point cloud from a file.

Computer Vision

import open3d as o3d

point_cloud = o3d.io.read_point_cloud([1])
print(point_cloud)

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

Apointcloud

B12345

CNone

D"data/pointcloud.pcd"

Attempts:

3 left

💡 Hint

Common Mistakes

Passing a number instead of a string path.

Forgetting to put quotes around the file path.

The read_point_cloud function requires a string path to the point cloud file.

2fill in blank

medium

Complete the code to create a 3D bounding box from a point cloud.

Computer Vision

bbox = point_cloud.get_axis_aligned_bounding_box()
print(bbox.[1]())

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

Aarea

Bvolume

Clength

Dcenter

Attempts:

3 left

💡 Hint

Common Mistakes

Using area() which is for 2D shapes.

Using center() which returns a point, not a size.

The volume() method returns the volume of the bounding box.

3fill in blank

hard

Fix the error in the code to visualize the point cloud with bounding box.

Computer Vision

import open3d as o3d

vis = o3d.visualization.Visualizer()
vis.create_window()
vis.add_geometry(point_cloud)
vis.add_geometry([1])
vis.run()
vis.destroy_window()

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

Abbox

BNone

Co3d.geometry.PointCloud()

Dpoint_cloud

Attempts:

3 left

💡 Hint

Common Mistakes

Adding the point cloud twice instead of the bounding box.

Adding None or creating a new empty geometry.

The bounding box object 'bbox' must be added to the visualizer to show it.

4fill in blank

hard

Fill both blanks to filter points inside the bounding box.

Computer Vision

points = np.asarray(point_cloud.points)
mask = (points[:, [1]] >= bbox.min_bound[[2]]) & (points[:, [1]] <= bbox.max_bound[[2]])
filtered_points = points[mask]

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

A0

B1

C2

D3

Attempts:

3 left

💡 Hint

Common Mistakes

Using different indices for points and bounding box bounds.

Using an invalid axis index like 3.

To filter along the x-axis, use index 0 for both point coordinates and bounding box bounds.

5fill in blank

hard

Fill all three blanks to train a simple 3D object detection model using PyTorch.

Computer Vision

import torch
import torch.nn as nn

class Simple3DDetector(nn.Module):
    def __init__(self):
        super().__init__()
        self.fc1 = nn.Linear([1], 64)
        self.relu = nn.ReLU()
        self.fc2 = nn.Linear(64, [2])

    def forward(self, x):
        x = self.relu(self.fc1(x))
        x = self.fc2(x)
        return x

model = Simple3DDetector()
input_tensor = torch.randn(10, [3])
output = model(input_tensor)
print(output.shape)

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

A3

B2

C5

D10

Attempts:

3 left

💡 Hint

Common Mistakes

Mismatching input tensor size with model input layer.

Wrong output size for classification.

Input features are 3 (x,y,z), output classes 2 (object or no object), input tensor batch size 10 with 3 features.