Edgar Rodríguez y Evelyn Orellana: Aplicaciones en IA con un enfoque exploratorio heurístico

Edgar Rodríguez y Evelyn Orellana: Aplicaciones en IA con un enfoque exploratorio heurístico

by Edgar Rodríguez García, Evelyn Orellana Orantes

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 Edgar Rodríguez García, Evelyn Orellana Orantes 👥 122 📅 November 13, 2025 ⏱ 73 min 👁 26 📄 expert opinion 🧭 2026-08-16
Available in: English (current) Français

Keywords

coin recognitionhand trackinghomomorphic encryptionMediaPipePaillier

Summary

This talk presents three exploratory AI projects developed by Edgar Rodríguez and Evelyn Orellana, co-founders of Revista Toolbar. The first project is the RODOR algorithm for identifying irregular surfaces with circular contours, applied to coin recognition and counting. It uses computer vision without neural networks, requiring only a single image of the coin’s face for training. The algorithm employs Hamming distance, Hough transform, and Canny edge detection, achieving 98.8% accuracy in tests with coins from multiple countries. The second project is real-time robotic simulation of human hand gestures, using a smartphone camera and MediaPipe Hand to track 21 landmarks, then mapping them to a virtual robotic hand in the CoppeliaSim simulator. The system works up to 5 meters away without specialized hardware. The third project is facial recognition for homomorphic encryption without key transport, applied to digital documents. It uses MTCNN for deterministic facial landmark extraction and Paillier homomorphic encryption to generate a digital signature dynamically, eliminating the need to store or transmit keys. The talk emphasizes a heuristic, exploratory approach, combining existing tools with custom algorithms to solve specific problems efficiently.

182 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the information lies in the presentation of three concrete, reproducible projects that demonstrate the application of AI techniques in practical scenarios. Each project is described with its heuristic components, methodology, and results, providing a clear understanding of the approach. The argumentation is solid, as the speakers justify their design choices and highlight the efficiency gains, such as using a single image for training in coin recognition and eliminating key transport in encryption. However, the talk is more of a showcase than a rigorous scientific presentation, lacking detailed statistical analysis or comparison with existing methods. The exploratory nature is acknowledged, and future work is suggested, which adds credibility.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is moderate: the projects are based on established techniques (e.g., Hough transform, MediaPipe, Paillier encryption) and the speakers provide references and source code links. However, the presentation does not include formal evaluation metrics beyond accuracy percentages, and the methodology is not fully detailed. The title accurately reflects the content, focusing on AI applications with an exploratory heuristic approach. No comments were provided, so no analysis of public trends is possible.

198 words

Title / Content Match

The title accurately reflects the content, which focuses on AI applications with an exploratory heuristic approach.

Quality & Reliability

7/10

The speakers are experienced professionals with advanced degrees in AI and computer science. They present three original exploratory projects with clear methodologies and results, but the presentation lacks formal peer review and some details are not fully elaborated.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • RODOR algorithm source code — Mentioned as source code for the coin recognition algorithm
  • MediaPipe Hand — Used for hand landmark detection in the simulation project
  • CoppeliaSim — Simulator used for the virtual robotic hand
  • Paillier cryptosystem — Homomorphic encryption scheme used in the facial recognition encryption project
  • MTCNN — Neural network for face detection used in the encryption project

Concurring Sources

  • MediaPipe Hands — Official documentation for the hand tracking solution used in the simulation.
  • Paillier cryptosystem — Wikipedia article explaining the Paillier encryption scheme.

Contribution & Novelties

The talk presents three original exploratory projects that combine existing AI techniques in novel ways. The RODOR algorithm offers a lightweight alternative to neural networks for specific object recognition tasks, achieving high accuracy with minimal training data. The hand gesture simulation demonstrates a low-cost, accessible approach to controlling virtual environments, potentially useful for education and accessibility. The facial recognition encryption system addresses a critical security vulnerability by eliminating key storage and transport, using biometric data as a dynamic key. These projects highlight the value of heuristic, exploratory research in AI.

Pour aller plus loin :

  • Hamming distance — Fundamental concept used in RODOR for comparing binary sequences.
  • Hough transform — Technique for detecting shapes, used in coin recognition.
  • Canny edge detector — Edge detection algorithm used in image processing.
  • Homomorphic encryption — Overview of encryption allowing computation on ciphertexts.
  • MediaPipe — Framework for building multimodal applied ML pipelines, used for hand tracking.

152 words

Radar Profile

The radar profile shows a balanced performance across all dimensions, with slightly higher scores in information quantity and reliability, reflecting the solid but not exceptional presentation. The technical level is moderate, suitable for a general technical audience.

Reliability 7/10