Buchendorfer Vision Contact

Computer vision prototyping · Aachen, Germany

Computer vision prototypes, tested before production integration.

We select the camera and lighting, prepare the data, develop and evaluate the model, and benchmark it on suitable edge hardware.

Discuss a project View RallyTrace
  • Imaging design
  • Model evaluation
  • Edge deployment
Illustrative machine vision camera above a machined component
Concept illustration of a prototype imaging setup.

01 / Services

Imaging, models and edge deployment.

Each project is scoped around one visual task and a defined test setup.

01

Imaging design

We select and test the camera, lens, sensor position and lighting for the feature that must be visible.

Deliverable
Camera and lighting specification
02

Model development and analysis

We prepare the data, train or adapt a suitable model, and evaluate missed and incorrect results.

Deliverable
Evaluated model and data workflow
03

Edge deployment

We run and benchmark the prototype on NVIDIA Jetson, an industrial PC or the customer’s network when appropriate.

Deliverable
Benchmarked edge prototype

Edge processing

The prototype can run near the camera.

When local processing is appropriate, we benchmark runtime and latency on the selected device and prepare the agreed interface.

  1. Camera and lighting
  2. Jetson or industrial PC
  3. API or prototype I/O
NVIDIA Jetson Orin Nano developer kit used for compact local AI processing
NVIDIA Jetson Orin Nano developer kit.

02 / Applications

Applications we can develop and test.

Feasibility depends on whether the relevant feature can be captured consistently and tested with representative data.

Illustrative computer vision prototype inspecting a machined component on a lab bench
Illustrative bench-scale inspection under controlled lighting.

01Industrial vision

Inspection and assembly checks

Test whether visible defects, missing parts, incorrect orientation or print differences can be detected under defined conditions.

Example output
Pass, review or reject output linked to a source image
Illustrative camera setup recording a tennis rally on a clay court
Illustrative court-side camera position for motion tracking.

02Industrial and sports vision

Object and motion tracking

Track defined objects, movements or events and retain links to the corresponding video frames.

Example output
Counts, events and linked footage
Illustrative engineer reviewing source frames beside a camera and edge computing prototype
Illustrative frame-review and edge-compute workspace.

03Visual data analytics

Image and video analysis

Extract measurements from images, video and model outputs, then present them in a review workflow.

Example output
Measurements and review interface

Medical R&D prototypes are engineering tools, not certified medical devices. Vision prototypes are not safety-rated control systems.

RallyTrace analysis over clay-court tennis footage
Original RallyTrace project footage.

03 / Bachelor’s project

RallyTrace tracks a tennis ball and links each result to the source video.

Jeremias developed the video-processing, tracking and review workflow. The project uses a lightweight TrackNet model, TensorRT acceleration and a shared frame timeline.

Tracking
Small, fast-object tracking
Runtime
TensorRT-accelerated inference
Review
Results linked to source frames
View the RallyTrace project
Engineering details

Adaptive search and tight crops help recover the ball in difficult frames. A lightweight TrackNet model runs with TensorRT, and outputs remain linked to the source-frame timeline.

04 / Process

A prototype project has three stages.

Before development starts, we agree the task, test conditions, deliverables and responsibility boundary.

  1. 01

    Scope the test

    Define the visual decision, representative data, required output and relevant error cases.

  2. 02

    Build the prototype

    Set up the imaging, data pipeline, model, analysis and selected edge device.

  3. 03

    Test and hand over

    Report the results and limitations, then provide the deliverables agreed in the project scope.

Buchendorfer Vision

Prototype and technical handover

A working prototype for the agreed conditions, imaging and compute specification, test results, and the agreed software and documentation.

Industrialization partner

Production integration

Final housing and mounting, plant electrical work, safety and conformity responsibilities, and permanent commissioning.

05 / Team

The founders carry out the project work.

Jeremias Buchendorfer

Co-founder · Computer vision, data and edge systems

Jeremias Buchendorfer

Computer vision, model development, data analytics and edge deployment.

Jeremias has worked on computer vision and data projects in industrial, medical and sports contexts, independently and within startup and established-company teams.

FocusComputer vision · Data analytics · Edge deployment · Prototyping

Co-founder · Electrical engineering

Sven Bordihn

Sven supports prototype hardware and electrical system-interface decisions.

FocusElectrical engineering · Prototype hardware · System interfaces

06 / Contact

Discuss a computer vision project.

In your first message, describe the object or process, the decision made today and the output you need. Sensitive footage is not required at this stage.

Send an email