A lightweight object detector for edge devices (Pattern Recognition 2023)
YOGA is a YOLO-style detector built for low-end hardware such as the NVIDIA Jetson Nano. Its CSPGhostNet backbone learns half of each layer’s feature maps with ordinary convolutions and generates the other half with cheap linear operations. Its AFF-PANet neck fuses multi-scale features with attentional feature fusion.
Compared with more than ten detectors on COCO, YOGA gives the best trade-off between size and accuracy: up to 22% higher AP with 23–34% fewer parameters and FLOPs. The four model sizes range from 1.9M parameters (32.3 AP) to 33.6M parameters (48.9 AP).
Code:GitHub. Pretrained COCO weights load in one line with torch.hub.load("TrustAIoT/YOGA_object-detection", "yoga_s"), and there is a Colab quick start.
YOGA is a deep-learning object detector built to run on low-end edge devices while staying accurate. Its backbone, CSPGhostNet, learns half of each layer’s feature maps with ordinary convolutions and produces the other half with cheap linear operations, which roughly halves the number of filters. Its neck, AFF-PANet, fuses features across scales with attentional feature fusion instead of plain concatenation. On COCO-val and COCO-testdev, compared with more than ten state-of-the-art detectors, YOGA gives the best trade-off between model size and accuracy: up to 22% higher AP with 23–34% fewer parameters and FLOPs. It also runs in real time on an NVIDIA Jetson Nano.
@article{sunkara2023yoga,title={{YOGA}: Deep Object Detection in the Wild with Lightweight Feature Learning and Multiscale Attention},author={Sunkara, Raja and Luo, Tie},journal={Pattern Recognition},volume={139},pages={109451},year={2023},publisher={Elsevier},doi={10.1016/j.patcog.2023.109451},}
