ghostjat/darknet

PHP bindings to the Darknet.

dev-main 2021-05-12 14:14 UTC

Requires

• php: ^7.4|^8.0
• ext-ffi: *
• cboden/ratchet: ^0.4.3

Apache-2.0

• Shubham Chaudhary

• php
• ai
• yolo
• ffi
• darknet
• yolov3
• yolov4
• yolo-tiny
• object-detection

README

[]

php-ffi experiment

php interface to the lib-darknet for object detection. Php7.4+ is required

Install

composer require ghostjat/darknet

Description

lib-darknet

Darknet

Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.

• It offers a simple API.
• High performance, due to the fact that it uses native interface elements.
• Fast learning by the user, due to the simplicity of its API.

yolo v2, v3, v4

If running on cpu you may use FASTEST cfg & weight for real time object detection on MS COCO dataset.

require '../vendor/autoload.php';

use darknet\core;
$dn = new core(core::YOLOFASTEST);

System-Conf:- CPU:- Intel(R) Core(TM) i3-2370M CPU @ 2.40GHz 64bit MEM:- 8GB Dataset:- MS-COCO Classes:- 80

YOLO	Time (ms)	CPU	Mem(mb)(max)
Tiny-v2	0.85933095 (max)	78	143.59
	0.62237596 (min)
Tiny-v3	0.93895602 (max)	90	125.9
	0.60306811 (min)
Main-v3	15.4672219 (max)	98	964.5
	14.0677847 (min)
Tiny-v4	0.85933095 (max)	82	151.9
	0.62237596 (min)
FASTEST	0.20039399 (max)	20	97.47
	0.11836814 (min)
FASTEST-xl	0.80017591 (max)	69	131.85
	0.22637511 (min)

Synopsis

WARNING:
This module is in its early stages and should be considered a Work in Progress. The interface is not final and may change in the future.

Sample:

Sample code:

require '../vendor/autoload.php';

use darknet\core;
$lib = '/home/ghost/bin/c-lib/darknet/data/';
$img = ['eagle.jpg','giraffe.jpg','horses.jpg','person.jpg','kite.jpg'];
$dn = new core();
foreach ($img as $value) {
    $dn->detect($lib.$value);
}

Author

Shubham Chaudhary [e-mail]