Authors: Xinhao Xiang, Simon Dräger, Jiawei Zhang

Published on: March 18, 2024

Impact Score: 7.6

Arxiv code: Arxiv:2403.12317

Summary

• What is new: EffiPerception framework that offers improved accuracy, speed, and memory usage across multiple computer vision tasks.
• Why this is important: Previous methods were often tailored to specific tasks or datasets, lacking versatility.
• What the research proposes: EffiPerception combines efficient feature extractors, layers, and an 8-bit optimizer to enhance performance across diverse perception tasks.
• Results: The framework outperformed previous methods on the KITTI, semantic-KITTI, and COCO datasets for detection and segmentation tasks.

Technical Details

Technological frameworks used: EffiPerception

Models used: Efficient Feature Extractors, Efficient Layers, EffiOptim 8-bit optimizer

Data used: KITTI, semantic-KITTI, COCO datasets

Potential Impact

Computer vision technology providers, automotive companies using 2D/3D object detection, and industries leveraging instance and point cloud segmentation could greatly benefit or be disrupted.

Want to implement this idea in a business?

We have generated a startup concept here: VisionAI.