import gradio as gr import os import requests from PIL import Image def face_compare(frame1, frame2): url = "https://faceapi.miniai.live/face_compare" files = {'file1': open(frame1, 'rb'), 'file2': open(frame2, 'rb')} r = requests.post(url=url, files=files) html = None faces = None compare_result = r.json().get('compare_result') compare_similarity = r.json().get('compare_similarity') html = ("" "" "" "" "" "" "" "" "" "" "" "" "" "
StateValue
Is same person? {compare_result}
Similarity{compare_similarity}
".format(compare_result=compare_result, compare_similarity=compare_similarity)) try: image1 = Image.open(frame1) image2 = Image.open(frame2) face1 = None face2 = None if r.json().get('face1') is not None: face = r.json().get('face1') x1 = face.get('x1') y1 = face.get('y1') x2 = face.get('x2') y2 = face.get('y2') if x1 < 0: x1 = 0 if y1 < 0: y1 = 0 if x2 >= image1.width: x2 = image1.width - 1 if y2 >= image1.height: y2 = image1.height - 1 face1 = image1.crop((x1, y1, x2, y2)) face_image_ratio = face1.width / float(face1.height) resized_w = int(face_image_ratio * 150) resized_h = 150 face1 = face1.resize((int(resized_w), int(resized_h))) if r.json().get('face2') is not None: face = r.json().get('face2') x1 = face.get('x1') y1 = face.get('y1') x2 = face.get('x2') y2 = face.get('y2') if x1 < 0: x1 = 0 if y1 < 0: y1 = 0 if x2 >= image2.width: x2 = image2.width - 1 if y2 >= image2.height: y2 = image2.height - 1 face2 = image2.crop((x1, y1, x2, y2)) face_image_ratio = face2.width / float(face2.height) resized_w = int(face_image_ratio * 150) resized_h = 150 face2 = face2.resize((int(resized_w), int(resized_h))) if face1 is not None and face2 is not None: new_image = Image.new('RGB',(face1.width + face2.width + 10, 150), (80,80,80)) new_image.paste(face1,(0,0)) new_image.paste(face2,(face1.width + 10, 0)) faces = new_image.copy() elif face1 is not None and face2 is None: new_image = Image.new('RGB',(face1.width + face1.width + 10, 150), (80,80,80)) new_image.paste(face1,(0,0)) faces = new_image.copy() elif face1 is None and face2 is not None: new_image = Image.new('RGB',(face2.width + face2.width + 10, 150), (80,80,80)) new_image.paste(face2,(face2.width + 10, 0)) faces = new_image.copy() except: pass return [faces, html] def check_liveness(frame): url = "https://faceapi.miniai.live/face_liveness_check" file = {'file': open(frame, 'rb')} r = requests.post(url=url, files=file) faceCount = None response_data = r.json() for item in response_data.get('face_state', []): if 'faceCount' in item: faceCount = item['faceCount'] break faces = None live_result = [] live_result.append(f"") for item in response_data.get('face_state', []): if item.get('FaceID'): faceID = item.get('FaceID') result = item.get('LivenessCheck') age = item.get('Age') gender = item.get('Gender') live_result.append(f"") live_result.append(f"
FaceIDAgeGenderLiveness
{faceID}{age}{gender}{result}
") live_result = ''.join(live_result) try: image = Image.open(frame) for face in r.json().get('faces'): x1 = face.get('x1') y1 = face.get('y1') x2 = face.get('x2') y2 = face.get('y2') if x1 < 0: x1 = 0 if y1 < 0: y1 = 0 if x2 >= image.width: x2 = image.width - 1 if y2 >= image.height: y2 = image.height - 1 face_image = image.crop((x1, y1, x2, y2)) face_image_ratio = face_image.width / float(face_image.height) resized_w = int(face_image_ratio * 150) resized_h = 150 face_image = face_image.resize((int(resized_w), int(resized_h))) if faces is None: faces = face_image else: new_image = Image.new('RGB',(faces.width + face_image.width + 10, 150), (80,80,80)) new_image.paste(faces,(0,0)) new_image.paste(face_image,(faces.width + 10, 0)) faces = new_image.copy() except: pass return [faces, live_result] def face_emotion(frame): url = "https://faceapi.miniai.live/face_emotion" file = {'file': open(frame, 'rb')} r = requests.post(url=url, files=file) emotion_result = [] emotion_result.append(f"") emotion_result.append(f"
Emotional Result : {r.json().get('emotion_result')}
") emotion_result = ''.join(emotion_result) faces = None try: image = Image.open(frame) for face in r.json().get('faces'): x1 = face.get('x1') y1 = face.get('y1') x2 = face.get('x2') y2 = face.get('y2') if x1 < 0: x1 = 0 if y1 < 0: y1 = 0 if x2 >= image.width: x2 = image.width - 1 if y2 >= image.height: y2 = image.height - 1 face_image = image.crop((x1, y1, x2, y2)) face_image_ratio = face_image.width / float(face_image.height) resized_w = int(face_image_ratio * 150) resized_h = 150 face_image = face_image.resize((int(resized_w), int(resized_h))) if faces is None: faces = face_image else: new_image = Image.new('RGB',(faces.width + face_image.width + 10, 150), (80,80,80)) new_image.paste(faces,(0,0)) new_image.paste(face_image,(faces.width + 10, 0)) faces = new_image.copy() except: pass return [faces, emotion_result] # APP Interface with gr.Blocks() as MiniAIdemo: gr.Markdown( """

FaceSDK Web Online Demo

Experience our NIST FRVT Top Ranked FaceRecognition, iBeta 2 Certified Face Liveness Detection Engine



""" ) with gr.Tabs(): with gr.Tab("Face Recognition"): with gr.Row(): with gr.Column(): im_match_in1 = gr.Image(type='filepath', height=300) gr.Examples( [ os.path.join(os.path.dirname(__file__), "images/compare/demo-pic22.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic60.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic35.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic33.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic34.jpg"), ], inputs=im_match_in1 ) with gr.Column(): im_match_in2 = gr.Image(type='filepath', height=300) gr.Examples( [ os.path.join(os.path.dirname(__file__), "images/compare/demo-pic41.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic32.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic39.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic61.jpg"), os.path.join(os.path.dirname(__file__), "images/compare/demo-pic40.jpg"), ], inputs=im_match_in2 ) with gr.Column(): im_match_crop = gr.Image(type="pil", height=256) txt_compare_out = gr.HTML() btn_f_match = gr.Button("Check Comparing!", variant='primary') btn_f_match.click(face_compare, inputs=[im_match_in1, im_match_in2], outputs=[im_match_crop, txt_compare_out]) with gr.Tab("Face Liveness Detection"): with gr.Row(): with gr.Column(scale=1): im_liveness_in = gr.Image(type='filepath', height=300) gr.Examples( [ os.path.join(os.path.dirname(__file__), "images/liveness/f_real_andr.jpg"), os.path.join(os.path.dirname(__file__), "images/liveness/f_fake_andr_mask3d.jpg"), os.path.join(os.path.dirname(__file__), "images/liveness/f_fake_andr_monitor.jpg"), os.path.join(os.path.dirname(__file__), "images/liveness/f_fake_andr_outline.jpg"), os.path.join(os.path.dirname(__file__), "images/liveness/f_fake_andr_outline3d.jpg"), os.path.join(os.path.dirname(__file__), "images/liveness/1.jpg"), os.path.join(os.path.dirname(__file__), "images/liveness/3.png"), os.path.join(os.path.dirname(__file__), "images/liveness/4.jpg"), ], inputs=im_liveness_in ) btn_f_liveness = gr.Button("Check Liveness!", variant='primary') with gr.Blocks(): with gr.Row(): with gr.Column(): im_liveness_out = gr.Image(label="Croped Face", type="pil", scale=1) with gr.Column(): livness_result_output = gr.HTML() btn_f_liveness.click(check_liveness, inputs=im_liveness_in, outputs=[im_liveness_out, livness_result_output]) with gr.Tab("Face Emotional Recognition"): with gr.Row(): with gr.Column(): im_emotion_in = gr.Image(type='filepath', height=300) gr.Examples( [ os.path.join(os.path.dirname(__file__), "images/emotion/1.jpg"), os.path.join(os.path.dirname(__file__), "images/emotion/2.jpg"), os.path.join(os.path.dirname(__file__), "images/emotion/3.jpg"), os.path.join(os.path.dirname(__file__), "images/emotion/4.jpg"), os.path.join(os.path.dirname(__file__), "images/emotion/5.jpg"), os.path.join(os.path.dirname(__file__), "images/emotion/6.jpg"), ], inputs=im_emotion_in ) btn_f_emotion = gr.Button("Check Emotion!", variant='primary') with gr.Blocks(): with gr.Row(): with gr.Column(): im_emotion_out = gr.Image(label="Result Image", type="pil", scale=1) with gr.Column(): txt_emotion_out = gr.HTML() btn_f_emotion.click(face_emotion, inputs=im_emotion_in, outputs=[im_emotion_out, txt_emotion_out]) if __name__ == "__main__": MiniAIdemo.launch()