GitHub PyPI - Downloads

DetExecutor

Python package with latest versions of YOLO architecture for training and inference

Install

Installing is quite simple, just use pip:

pip3 install det_executor

Train

Training support is still in progress!

Inference

Get available models

from det_executor import DetExecutor
# print list of supported arches
DetExecutor.list_arch()

Output

{
    "yolov7": YoloArch(
        version="7",
        img_size=(640, 640),
        size="75.6MB",
        params="37.6M",
        flops="",
        module="yolov7_package",
        load_link="yolov7.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7x": YoloArch(
        version="7",
        img_size=(640, 640),
        size="75.6MB",
        params="71.3M",
        flops="",
        module="yolov7_package",
        load_link="yolov7x.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7-w6": YoloArch(
        version="7",
        img_size=(1280, 1280),
        size="141.3MB",
        params="70.4M",
        flops="",
        module="yolov7_package",
        load_link="yolov7-w6.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7-e6": YoloArch(
        version="7",
        img_size=(1280, 1280),
        size="195.0MB",
        params="97.2M",
        flops="",
        module="yolov7_package",
        load_link="yolov7-e6.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7-d6": YoloArch(
        version="7",
        img_size=(1280, 1280),
        size="286.3MB",
        params="133.8M",
        flops="",
        module="yolov7_package",
        load_link="yolov7-d6.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7-e6e": YoloArch(
        version="7",
        img_size=(1280, 1280),
        size="304.4MB",
        params="151.8M",
        flops="",
        module="yolov7_package",
        load_link="yolov7-e6e.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7-traced": YoloArch(
        version="7",
        img_size=(640, 640),
        size="74.3MB",
        params="36.9M",
        flops="",
        module="yolov7_package",
        load_link="1L8mPcUvabUscEk6Nr8ck5EFgopgPAMDW",
        trainable=False,
        traced=True,
    ),
    "yolov7-tiny": YoloArch(
        version="7",
        img_size=(640, 640),
        size="12.6MB",
        params="6.2M",
        flops="",
        module="yolov7_package",
        load_link="yolov7-tiny.pt",
        trainable=False,
        traced=False,
    ),
    "yolov7-tiny-traced": YoloArch(
        version="7",
        img_size=(640, 640),
        size="12.7MB",
        params="6.2M",
        flops="",
        module="yolov7_package",
        load_link="18zJyljtolPENDI_kFw3FlRFnQTnaLuDF",
        trainable=False,
        traced=True,
    ),
    "yolov8n": YoloArch(
        version="8",
        img_size=(640, 640),
        size="6.5MB",
        params="3.2M",
        flops="",
        module="yolov8",
        load_link="yolov8n.pt",
        trainable=False,
        traced=False,
    ),
    "yolov8s": YoloArch(
        version="8",
        img_size=(640, 640),
        size="22.6MB",
        params="11.2M",
        flops="",
        module="yolov8",
        load_link="yolov8s.pt",
        trainable=False,
        traced=False,
    ),
    "yolov8m": YoloArch(
        version="8",
        img_size=(640, 640),
        size="52.1MB",
        params="25.9M",
        flops="",
        module="yolov8",
        load_link="yolov8m.pt",
        trainable=False,
        traced=False,
    ),
    "yolov8l": YoloArch(
        version="8",
        img_size=(640, 640),
        size="87.8MB",
        params="43.7M",
        flops="",
        module="yolov8",
        load_link="yolov8l.pt",
        trainable=False,
        traced=False,
    ),
    "yolov8x": YoloArch(
        version="8",
        img_size=(640, 640),
        size="136.9MB",
        params="68.2M",
        flops="",
        module="yolov8",
        load_link="yolov8x.pt",
        trainable=False,
        traced=False,
    ),
    "yolos-tiny": YoloArch(
        version="s",
        img_size=None,
        size="136.9MB",
        params="6.5M",
        flops="512x*>18.8G|256x*>3.4G",
        module="yolos",
        load_link="hustvl/yolos-tiny",
        trainable=False,
        traced=False,
    ),
}

Loading model

from det_executor import DetExecutor

# loading model
name = 'yolov7'
ex = DetExecutor(name)

Predict and draw

from det_executor import DetExecutor, draw_on_image
import cv2

# loading model
name = 'yolov7'
ex = DetExecutor(name)

# loading image
img = ex.load_image('test/img.jpg')
# or img = cv2.imread('test/img.jpg')

# predict
classes, boxes, scores = ex.predict(img)

# draw
img = draw_on_image(img, boxes[0], scores[0], classes[0])
cv2.imshow("image", img)
cv2.waitKey()

Roadmap

Training pipeline for all models
Load from custom weights
More models

Citation

@article{wang2022yolov7,
  title={{YOLOv7}: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors},
  author={Wang, Chien-Yao and Bochkovskiy, Alexey and Liao, Hong-Yuan Mark},
  journal={arXiv preprint arXiv:2207.02696},
  year={2022}
}

@misc{fang2021look,
      title={You Only Look at One Sequence: Rethinking Transformer in Vision through Object Detection}, 
      author={Yuxin Fang and Bencheng Liao and Xinggang Wang and Jiemin Fang and Jiyang Qi and Rui Wu and Jianwei Niu and Wenyu Liu},
      year={2021},
      eprint={2106.00666},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

License
- OSI Approved :: MIT License
Operating System
- OS Independent
Programming Language
- Python :: 3.6

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

det_executor-0.0.9.tar.gz (97.2 kB view details)

Uploaded Oct 6, 2023 Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

det_executor-0.0.9-py3-none-any.whl (103.2 kB view details)

Uploaded Oct 6, 2023 Python 3

File details

Details for the file det_executor-0.0.9.tar.gz.

File metadata

Download URL: det_executor-0.0.9.tar.gz
Upload date: Oct 6, 2023
Size: 97.2 kB
Tags: Source
Uploaded using Trusted Publishing? No
Uploaded via: twine/4.0.2 CPython/3.11.0

File hashes

Hashes for det_executor-0.0.9.tar.gz
Algorithm	Hash digest
SHA256	`bf91cb2371d5e6c2d67646c4eb5483853b7f571e9da5bff6896986b8eb508667`
MD5	`fcaf0c97bb3c8a69a2fc0649b2b9d5fc`
BLAKE2b-256	`7b4faa99baf2d68347ec3a01f1a6289f14eb8d27f16b70bda20be79139df1d78`

See more details on using hashes here.

File details

Details for the file det_executor-0.0.9-py3-none-any.whl.

File metadata

Download URL: det_executor-0.0.9-py3-none-any.whl
Upload date: Oct 6, 2023
Size: 103.2 kB
Tags: Python 3
Uploaded using Trusted Publishing? No
Uploaded via: twine/4.0.2 CPython/3.11.0

File hashes

Hashes for det_executor-0.0.9-py3-none-any.whl
Algorithm	Hash digest
SHA256	`b36858a2a60337e048eb396ba8f98c160b497177dfa31d31474a2c5a4adaf79c`
MD5	`d71816de46e8bce14ac6031589efa63a`
BLAKE2b-256	`38b76726bb3c2c9b61ea911977dfc0dc283cfc84a20a22b5b0962bb06d8702ac`