Depth Identification Network for Vehicle Image Distance Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for depth identification in vehicle images fail to accurately determine distances between vehicles and surrounding objects or obstacles, affecting driving safety due to the inability to extract depth information through image recognition models.
Innovation Solution
A method involving a computer device that uses a camera to capture images, obtains point clouds and spatial coordinates, and applies a depth identification network to generate initial and target depth images, adjusting the network based on loss values to create a pre-trained model capable of accurately determining depth information with measurement units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current image recognition models are used for depth identification, then the model structure is simple and easy to implement, but the depth information cannot be accurately identified and measurement units are lost
Solution Approach 1:
The patent merges the depth identification network with the image recognition model, integrating depth estimation functionality into the existing model architecture. This allows the model to simultaneously perform image recognition and depth identification, resolving the contradiction by combining multiple functions into a unified system that maintains relative simplicity while achieving accurate depth measurement with measurement units
Solution Approach 2:
The patent introduces a new dimensional output for depth information with measurement units alongside the traditional image recognition outputs. By adding this measurement dimension to the model's output space, the system achieves precise depth identification without fundamentally restructuring the entire model, thus resolving the contradiction between measurement precision and model complexity
2Reliability
If depth information is not accurately identified, then the image recognition model remains simple, but driving safety is compromised due to inability to determine accurate distances
Solution Approach 1:
The patent implements a feedback mechanism where the depth identification network provides depth information with measurement units back to the image recognition system. This feedback loop enables the system to continuously refine distance measurements and improve driving safety decisions based on accurate depth information, resolving the contradiction by using feedback to enhance reliability without sacrificing measurement precision
Data Source
AI summary
A method for identifying depths of images is provided. In the method, the computer device obtains point clouds of a road scene and a spatial coordinate value of each point in the point clouds. An initial depth image is obtained by inputting the first image into a preset depth identification network. A projected depth value and a projected coordinate value are obtained by converting the spatial coordinate value, and a target depth value is calculated. An initial projection image is generated based on the target depth value and a second image. A loss value is calculated according to the first image, the initial projection image and the second image, and a pre-trained image identification model is obtained by adjusting the preset depth identification network. By performing the method, measurement unit information of depth information of images can be determined.