Equicares

The problem

Create and track equestrian sessions in a management application, then provide posture observations from images. Detecting joint keypoints is not enough: they need to be presented in a clear user flow, and the feedback given to the rider needs evaluation.

My role

I designed management workflows and developed the SaaS frontend and backend. I integrated YOLO for pose analysis and explored angle visualization and position feedback.

Constraints

Connect technical choices to the needs of the equestrian domain.
Build the complete application and its business interactions.
Turn model output into useful information for the user.
Distinguish implemented features from exploratory ideas.

Architecture & pipeline

Conceptual diagram

Design and develop the management web application.
Integrate YOLO to analyze rider pose from an image.
Explore posture visualization and position feedback.
Investigate LLM-based interpretation, advice and exercise suggestions.

Decisions & iterations

I built session creation and tracking workflows before integrating image analysis.

YOLO keypoints are used to compute and display posture angles; feedback quality needs evaluation separately from detection.

LLM-based interpretation, advice and exercise generation remain exploratory, without validation for real use.

Results & limitations

An equestrian management application developed end to end, with session creation and tracking.

Exploratory rider pose analysis with YOLO and angle visualization from detected keypoints.

Posture feedback still needs evaluation. LLM recommendations and exercise generation are not validated features.

Next iterations

Document use cases and posture analysis limitations. Evaluate feedback quality before extending the advice features.

In pictures

The mobile flow for creating a session and recording observations.

Exploratory analysis of rider posture and joint angles — Exploration of posture feedback from detected keypoints. Advice quality still needs evaluation.