Method of pre-fusion detection model based on panoramic camera
CN119007147APending Publication Date: 2024-11-22CHANGAN AUTOMOBILE (GRP) CO LTD
Patent Information
- Application Number
- CN202411030989.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-22
Smart Images

Figure CN119007147A_ABST
Abstract
A method of a front fusion detection model based on panoramic cameras comprises the following steps: 1) obtaining a plurality of panoramic images of the surrounding environment of an unmanned vehicle body through a plurality of panoramic cameras, and inputting the plurality of panoramic images into a convolutional neural network model; 2) the convolutional neural network model extracts image feature maps and feature offsets of all panoramic images in sequence; 3) shifting the image feature map of each panoramic image according to the feature offset of each panoramic image to form a shifting feature map of the image feature map of each panoramic image, and mapping a BEV feature map of each panoramic camera according to the shifting feature map of the image feature map of each panoramic image; and 4) stacking the BEV feature maps of the panoramic cameras along the channel direction, carrying out convolution on the stacked BEV feature maps, and outputting a target feature map after multi-view fusion.
Need to check novelty before this filing date? Find Prior Art
Citation Information
Patent Citations
Around view provision apparatus and vehicle including the same
CN105313779A
Transform-based high-precision map real-time prediction method and system
CN116071721A
Deep learning algorithm for multi-camera 3D target detection based on BEV
CN116385845A
High-precision map construction method and system based on laser radar and camera fusion
CN117008150A
Multi-view 3D target detection method based on sampling adaptive continuous NeRF
CN118071999A
