Frequently Asked Questions
What clubs, academies and coaches ask most often before onboarding.
What exactly does TechForSport do?
TechForSport is sports analysis software that turns training footage shot with a single camera into measured performance data. It extracts skeletal and posture data from the footage through 2D pose estimation, tracks players and the ball, and tags events. Those signals become five measurement categories: sprint split times at 10, 20 and 30 metres for speed; change-of-direction deficit (CODD) for agility, meaning the seconds lost relative to a straight sprint; aerobic speed reached in the 30-15 IFT test (VIFT) for endurance; barbell velocity for strength work; and jump height with flight time for power. The output is a coach report, compared session to session across 7, 28 and 90 day windows. Processing time is measured, not promised: a two-minute clip is processed in under a minute, so a session report is ready while the team is still in the locker room. Football and tennis run as field pilots today; volleyball, basketball and golf are in development.
Do we need sensors, wearables or a special camera?
No. The system works with footage from an ordinary camera, and a phone camera is enough. No athlete-worn sensors, GPS vests, markers or pitch-side calibration hardware are required, because the measurement comes from the footage itself rather than from a device. What matters instead is capture conditions, and we measured these rather than guessing them. The athlete should fill at least 25% of the frame height: in our earliest field footage they filled 6%, and measurement is not reliable at that scale. Tests that derive height from flight time require 60 frames per second; at 30 fps the error doubles on a 40 centimetre jump. In sprint and change-of-direction tests the distance must be measured on the field, because a wrong distance once produced an impossible 51 km/h reading. The camera must stay still, since handheld footage shifts the ground reference the measurement depends on.
Which sports are supported?
Football and tennis are in active field pilots today: the football pilot runs with Düzcespor and the tennis pilot with Tekirdağ Maris Tenis Spor Kulübü. Volleyball, basketball and golf are in development, and the interface labels that status explicitly rather than implying availability. All five share the same vision core; what differs per sport is the event vocabulary and the metric definitions. Football emphasises automatic tagging, pass and shot analysis, sprint and pressing patterns; tennis emphasises stroke classification, court positioning, serve placement and rally patterns. The athletic measurement layer, however, is sport-independent: sprint split times, change-of-direction deficit, aerobic speed, barbell velocity and jump height are measured the same way whichever sport you work in. Adding a new sport therefore does not mean rebuilding the vision engine — it means writing that sport's event vocabulary and metric definitions on top of the existing one.
How long does processing take?
A two-minute clip is processed in under a minute. We measured this rather than estimating it: the pipeline runs at roughly 0.4 times real time, so footage is analysed faster than its own duration. In practice this means a training session report is ready while the team is still in the locker room, and the coach can give feedback without waiting for the next session. The figure depends on the hardware doing the work, and we took this measurement on CPU, so no dedicated graphics card is required to reach it. Real-time analysis of a live camera feed is not covered by this measurement and remains a separate decision. Because raw footage is deleted once measurements have been extracted, processing happens once: the same clip cannot be re-analysed later, which is why getting capture conditions right the first time matters.
Where is athlete data processed and stored?
Athlete data is processed and stored inside the European Union, with the database in the AWS Frankfurt region. The region was not incidental — it was chosen deliberately for GDPR and Turkish KVKK Article 9 reasons. What is retained and for how long is published in the privacy policy, and the code uses the same values as that commitment: audit records are deleted automatically after 12 months and demo requests after 24 months. Raw footage is never archived; it is deleted once measurements have been extracted, leaving only the measurement results and the video record. Email is sent through Ireland and hosting runs from a European region. Endpoints that return personal data are never cached. When a subscription ends there is a 30-day window to export data before it is deleted from the systems. Sub-processors and processing terms are set out in the data processing agreement.
Is raw video kept?
No. Raw footage is deleted once measurements have been extracted, and the system keeps no permanent video archive. What remains is the measurement results plus the video record: which athlete, which test, which date, which duration. This is an architectural decision rather than a preference, and it has two consequences. The first is privacy: athlete footage, particularly of players under 18, is the most sensitive data category in the product, and data that is not stored cannot be leaked. The second is a practical limit: an uploaded clip cannot be re-processed later, because it no longer exists. Re-queueing a failed analysis job by hand achieves nothing; the session has to be filmed again. That is why capture conditions matter from the start — frame fill, frame rate and a still camera. A permanent video archive would be a separate decision and a separate consent question.
How does it work for athletes under 18?
Footage of an athlete under 18 cannot be processed without parental consent, and the software enforces this rather than leaving it to the operator. If the date of birth is unknown the athlete is treated as a minor, so the safe side is the default under uncertainty. The rule is written as pure functions and runs identically on the server and in the interface, which means bypassing the interface and calling the API directly is not a way around it. Edge cases are protected by their own test file, so if someone later loosens the rule by accident, the test fails. In academy and federation sales this is usually the first objection raised, and the answer is not a policy paragraph but a constraint enforced in the system. The consent record stays visible on the athlete card, so which footage was processed under which consent can be audited afterwards.
Do you measure force and power?
We do not measure force in newtons. Force and power cannot be measured directly from a single camera; measuring them requires a force plate. Test types whose output is force — Nordic hamstring, isometric tests, static single-leg variants — were therefore left outside the measurement scope rather than estimated and presented as measurements. What we measure in strength training is velocity: barbell velocity, velocity loss within a set as a percentage (%), and the load-velocity profile. These can be measured reliably from footage and the literature is mature; published work reports agreement of r = 0.940 between camera-based barbell velocity and reference systems (Balsalobre-Fernández 2020). That figure describes the accuracy of the method in the literature, not our own validation result. Heart-rate-based metrics are also out of scope, because a camera cannot measure them reliably. The distinction between what is measured and what is derived by model is recorded for every metric.
Do you predict injuries?
No. The system shows the coach how measurements change; it does not forecast an outcome. What it surfaces is concrete: left-right asymmetry in single-leg jumps, velocity loss within a set as a percentage, how sprint split times trend against previous tests, whether change-of-direction deficit differs between the right and left turn. None of that is a diagnosis, a prediction, or a promise to prevent injury, and it is not presented as one. The decision belongs to the coaching and medical staff; the software's job is to show the data accurately and in time. We draw this line deliberately, because any claim about injury incidence is a clinical claim and cannot be supported by single-camera measurement. Comparison windows are 7, 28 and 90 days, since an athlete's change against their own history means more than a ranking against other athletes.
Does it replace our existing analysis tools?
Not necessarily. The goal is to unify athletic measurements with technical and team data in a single athlete profile, and clubs can run it alongside the tools they already have. What sets it apart is that obtaining the data requires no new hardware and no separate analyst team: you do not need a timekeeper to get a split time from a sprint, a force plate to get height from a jump, or a vest to get aerobic speed from a run. Measurements from the same footage are then aggregated to answer three different questions: an athlete sees their own progress, a coach sees the fault a group shares, a club sees an age-group profile and its change across the season. The measurement is identical; only the level of the question changes. During onboarding, which tests run on which schedule is configured together rather than imposed as a template.
How do we get started?
The product is currently invite-only: accounts are opened after a demo call. The reason is that each club's measurement protocol and consent setup are configured together during onboarding — which tests will run, on what schedule, and how parental consent is collected for athletes under 18. Requests can be made through the demo form on the home page or by email. In the first conversation we usually suggest working with footage filmed on your own pitch, because seeing the value of the measurement in your own athletes' data is more convincing than a prepared example. Filming needs no special equipment: a phone camera, a stable tripod and the right angle are enough, with the capture conditions described elsewhere on this page. After onboarding, data export and account deletion are available to users, so starting is not an irreversible commitment.
Can we export or delete our data?
Yes. Data export and account deletion are available to users from within the account; these correspond to your rights under GDPR and Turkish KVKK Article 11, and they do not require filing a request and waiting. After a subscription ends there is a 30-day window to export your data, after which it is deleted from the systems; copies in backups fall away as the backup cycle completes. Because raw footage is already deleted once measurements have been extracted, what you export is measurement results, athlete records and reports rather than video. Audit records are deleted automatically after 12 months and demo requests after 24 months, and those retention periods are commitments published in the privacy policy rather than internal defaults. The controller-processor terms, the list of sub-processors and the deletion obligations are set out in the data processing agreement.
Which company is TechForSport?
TechForSport is the product of a sports technology company based in Türkiye, registered as T4F Spor Teknolojileri ve Yazılım A.Ş.. The company is established in Düzce, its Turkish trade registry (MERSIS) number is 0815120376200001, and its only official domain is t4fsport.com. It is not affiliated with any similarly named company.
Question you could not find an answer to?