๐Ÿ” HyperVision โ€” Object Detection

Premium Edition by Myndlabs.tech  ยท  Lightweight anchor-free object detection

๐Ÿ“‹ Overview

HyperVision is a lightweight anchor-free object detection model, evolved from the NanoDet-Plus architecture. It delivers real-time detection on mobile and edge devices while maintaining competitive accuracy. The model uses Generalized Focal Loss with a ShuffleNetV2 backbone and Ghost-PAN feature pyramid.

object-detection computer-vision lightweight mobile pytorch ncnn mnn openvino onnx

๐Ÿ—๏ธ Architecture

TypeAnchor-free one-stage detector (FCOS-style)
LossGeneralized Focal Loss (QFL + DFL + GIoU)
BackboneShuffleNetV2 (configurable)
NeckGhost-PAN feature pyramid
HeadNanoDet-Plus head with AGM & DSLA

๐Ÿš€ Deployment Backends

ncnnMobile / Android C++
MNNMobile / Embedded C++
OpenVINOIntel CPU / GPU
ONNXCross-platform
LibTorchC++ inference
PyTorchPython inference

๐Ÿง  Supported Backbones

ShuffleNetV2, ResNet, MobileNetV2, EfficientNet-Lite, GhostNet, RepVGG, Custom CSPNet, TIMM models

๐ŸŽฏ Try It Yourself

To run inference with HyperVision, you can use the GitHub repository or the Marimo notebook below.

๐Ÿ“– Citation

@misc{hypervision,
    title={HyperVision: Lightweight anchor-free object detection model},
    author={Yethikrishna R},
    howpublished={\url{https://github.com/Yethikrishna/hypervision}},
    year={2025},
    note={Premium edition published by Myndlabs.tech}
}