emYt+

Video conference compliance software.

TensorGo introduces emYt+, a low-code Platform as a Service (PaaS) designed to transform the way enterprises handle compliance and decision-making processes. Leveraging state-of-the-art deep learning technologies, emYt+ specializes in analyzing videos and images to detect fraudulent activities, particularly valuable in interviews and investigations. This intelligent assistant tool is tailor-made for HR professionals and investigation agencies, offering seamless integrations with popular platforms like Zoom and Webex. With its advanced features ranging from voice biometrics to emotion detection, emYt+ promises to enhance accuracy, efficiency, and security in enterprise operations, making deep learning solutions more accessible and effective for businesses of all sizes.

emyt+ dashboard.

</div>

Algorithms in emYt+

emYt+ leverages various machine learning and computer vision algorithms to provide a comprehensive video conference compliance software. Here's a look at some of the key algorithms and techniques applied:

  • Facial Emotion Recognition: Utilizing OpenFace for facial action unit detection and CNNs for emotion classification.
  • Demographics Estimation: Deep learning models estimate age, gender, and ethnicity.
  • Head Pose Estimation: SVM Regression Models for pitch, yaw, and roll orientation.
  • Heart Rate Estimation: Signal processing from facial landmarks.
  • Eye Gaze Estimation: Tracking and analyzing the movement of the eyes to infer focus and attention.
  • Lip Sync Analysis: Assessing the synchronization of speech and lip movement for accurate speech recognition.
  • Voice Diarization: Distinguishing between speakers in the video to provide speaker-specific transcriptions.
  • Voice Biometrics: Using unique voice features for speaker identification and authentication.
  • Speech Processing: ASR and Speaker Diarization to attribute transcriptions to individual speakers.

emYt+ Tech Stack (2020-2022)

  • Operating System: Ubuntu
  • Programming Languages: Python3, C++
  • Machine Learning Frameworks:
    • PyTorch
    • TensorFlow
    • Keras
    • numpy
    • Scikit-learn
    • Optuna
  • Computer Vision Modules:
    • OpenFace
    • Cv2 (computer vision module)
  • GPU Acceleration:
    • CUDA
    • cuDNN
  • Containerization:
    • Docker
    • Oracle Docker Registry
  • Multimedia Framework: GStreamer
  • WebRTC Media Server: Janus
  • Inference Servers:
    • Nvidia Triton-Server
    • TensorRT
  • Streaming Protocol: RTMP
  • Media Server: Node Media Server
  • Model Optimization: OpenVINO
  • Databases:
    • Oracle
    • InfluxDB
    • Object Storage (s3fs)
  • Monitoring and Visualization: Grafana
  • Enterprise Platform: Kubernetes
  • Messaging and Streaming: Kafka
  • API and Documentation:
    • FastAPI
    • Swagger
  • Continuous Integration/Continuous Deployment: CI/CD Pipelines

References