yolov7-onnx-infer

Inference with yolov7's onnx model

onnx model

• yolov7-tiny.onnx
• yolov7.onnx

1. Dev Env

1.1 pip install

pip install -U pip && pip install -r requirements.txt

1.2 Docker

Dockerfile

Ubuntu 20.04

FROM ubuntu:20.04
USER root

RUN ln -sf /usr/share/zoneinfo/Asia/Tokyo /etc/localtime

LABEL version="1.0"
LABEL description="Build operating environment for yolov7-onnx-infer"

RUN apt-get update && \
    apt-get -y install python3-pip && \
    apt-get -y install git && \
    apt-get -y install libgl1-mesa-dev && \
    apt-get -y install libglib2.0-0 && \
    pip install -U pip && \
    pip install opencv-python onnxruntime

Debian

FROM debian:stable-slim
USER root

LABEL version="1.0"
LABEL description="Build operating environment for yolov7-onnx-infer"

RUN apt-get update && \
    apt-get -y install python3-pip && \
    apt-get -y install git && \
    apt-get -y install libgl1-mesa-dev && \
    apt-get -y install libglib2.0-0 && \
    pip install -U pip && \
    pip install opencv-python onnxruntime

docker build

docker build -t tatsuya060504/yolov7-onnx-infer:v1.0.0 .

docker pull

Docker Hub

docker pull tatsuya060504/yolov7-onnx-infer:raspberrypi

docker run

docker run -it --name=yolov7-onnx-infer -v $(pwd):/home tatsuya060504/yolov7-onnx-infer:raspberrypi

2. Onnx Model Download

You can download the yolox-onnx model by executing the shell script in the model folder.

cd model
sh download_yolov7_tiny_onnx.sh #or download_yolov7_onnx.sh

3. Inference

Image

python onnx_inference.py -i <image_path> -m <onnx_model_path> -e <class_name> -s <score_threshold>

Video

python onnx_inference.py -mo video -i <video_path> -m <onnx_model_path> -e <class_name> -s <score_threshold>

4. References

• yolov7
• yolov7/tools/YOLOv7onnx.ipynb

5. About me

• Blog
• Twitter
• Youtube