YOLOv5

Ultralytics YOLOv5 model in Pytorch.

Proof of concept for (TypoSquatting, Niche Squatting) security flaw on Hugging Face.

Model Description

How to use

from transformers import YolosFeatureExtractor, YolosForObjectDetection
from PIL import Image
import requests

url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
image = Image.open(requests.get(url, stream=True).raw)

feature_extractor = YolosFeatureExtractor.from_pretrained('mhyatt000/yolov5')
model = YolosForObjectDetection.from_pretrained('mhyatt000/yolov5')

inputs = feature_extractor(images=image, return_tensors="pt")
outputs = model(**inputs)
# model predicts bounding boxes and corresponding COCO classes

logits = outputs.logits
bboxes = outputs.pred_boxes

Training Data

Training

Evaluation

Model was evaluated on COCO2017 dataset.

Model	size (pixels)	mAPval	Speed	params	FLOPS
YOLOv5s6	1280	43.3	4.3	12.7	17.4
YOLOv5m6	1280	50.5	8.4	35.9	52.4
YOLOv5l6	1280	53.4	12.3	77.2	117.7
YOLOv5x6	1280	54.4	22.4	141.8	222.9

Bibtex and citation info

@software{glenn_jocher_2022_6222936,
  author       = {Glenn Jocher and
                  Ayush Chaurasia and
                  Alex Stoken and
                  Jirka Borovec and
                  NanoCode012 and
                  Yonghye Kwon and
                  TaoXie and
                  Jiacong Fang and
                  imyhxy and
                  Kalen Michael and
                  Lorna and
                  Abhiram V and
                  Diego Montes and
                  Jebastin Nadar and
                  Laughing and
                  tkianai and
                  yxNONG and
                  Piotr Skalski and
                  Zhiqiang Wang and
                  Adam Hogan and
                  Cristi Fati and
                  Lorenzo Mammana and
                  AlexWang1900 and
                  Deep Patel and
                  Ding Yiwei and
                  Felix You and
                  Jan Hajek and
                  Laurentiu Diaconu and
                  Mai Thanh Minh},
  title        = {{ultralytics/yolov5: v6.1 - TensorRT, TensorFlow 
                   Edge TPU and OpenVINO Export and Inference}},
  month        = feb,
  year         = 2022,
  publisher    = {Zenodo},
  version      = {v6.1},
  doi          = {10.5281/zenodo.6222936},
  url          = {https://doi.org/10.5281/zenodo.6222936}
}