app.py

# SPDX-License-Identifier: MIT

import cv2
import onnxruntime as ort
import argparse
import numpy as np
from dependencies.box_utils import predict

# ------------------------------------------------------------------------------------------------------------------------------------------------
# Face detection using UltraFace-320 onnx model
os.system("wget https://github.com/AK391/models/raw/main/vision/body_analysis/ultraface/models/version-RFB-320.onnx")
face_detector_onnx = "version-RFB-320.onnx"

# Start from ORT 1.10, ORT requires explicitly setting the providers parameter if you want to use execution providers
# other than the default CPU provider (as opposed to the previous behavior of providers getting set/registered by default
# based on the build flags) when instantiating InferenceSession.
# For example, if NVIDIA GPU is available and ORT Python package is built with CUDA, then call API as following:
# ort.InferenceSession(path/to/model, providers=['CUDAExecutionProvider'])
face_detector = ort.InferenceSession(face_detector_onnx)

# scale current rectangle to box
def scale(box):
    width = box[2] - box[0]
    height = box[3] - box[1]
    maximum = max(width, height)
    dx = int((maximum - width)/2)
    dy = int((maximum - height)/2)

    bboxes = [box[0] - dx, box[1] - dy, box[2] + dx, box[3] + dy]
    return bboxes

# crop image
def cropImage(image, box):
    num = image[box[1]:box[3], box[0]:box[2]]
    return num

# face detection method
def faceDetector(orig_image, threshold = 0.7):
    image = cv2.cvtColor(orig_image, cv2.COLOR_BGR2RGB)
    image = cv2.resize(image, (320, 240))
    image_mean = np.array([127, 127, 127])
    image = (image - image_mean) / 128
    image = np.transpose(image, [2, 0, 1])
    image = np.expand_dims(image, axis=0)
    image = image.astype(np.float32)

    input_name = face_detector.get_inputs()[0].name
    confidences, boxes = face_detector.run(None, {input_name: image})
    boxes, labels, probs = predict(orig_image.shape[1], orig_image.shape[0], confidences, boxes, threshold)
    return boxes, labels, probs

# ------------------------------------------------------------------------------------------------------------------------------------------------
# Main void

def inference(img):
  color = (255, 128, 0)
  
  orig_image = cv2.imread(img)
  boxes, labels, probs = faceDetector(orig_image)
  
  for i in range(boxes.shape[0]):
      box = scale(boxes[i, :])
      cv2.rectangle(orig_image, (box[0], box[1]), (box[2], box[3]), color, 4)
      cv2.imwrite("out.png",orig_image)
  return "out.png"
  

gr.Interface(inference,gr.inputs.Image(type="filepath"),gr.outputs.Image(type="file")).launch()