Create AI CM:sport task

This algorithm allows to identify various sporting activities in a video. Generic info about all capabilities and limits see in the generic “Content Moderation” method.

What is Sports activity detection?

Sports activity detection by AI involves using machine learning and computer vision technologies to automatically identify, analyze, and interpret various activities within sports and generic videos. This can include detecting specific types, actions, events, and moments.

This model operates on a video sequence (and not on images as most of the used computer vision models). Make sure your video is at least 10-15 seconds long.

Sports activities can be detected:

• archery
• arm wrestling
• playing badminton
• playing baseball
• basketball dunk
• bowling
• boxing punch
• boxing speed bag
• catching or throwing baseball
• catching or throwing softball
• cricket
• curling
• disc golfing
• dodgeball
• fencing
• football
• golf chipping
• golf driving
• golf putting
• hitting baseball
• hockey stop
• ice skating
• javelin throw
• juggling soccer ball
• kayaking
• kicking field goal
• kicking soccer ball
• playing cricket
• playing field hockey
• playing ice hockey
• playing kickball
• playing lacrosse
• playing ping pong
• playing polo
• playing squash or racquetball
• playing tennis
• playing volleyball
• pole vault
• riding a bike
• riding or walking with horse
• roller skating
• rowing
• sailing
• shooting goal (soccer)
• skateboarding
• skiing

Use cases:

• Sports leagues and content creators can use AI to monitor UGC for unauthorized publications of their content. This can include detecting specific sporting events or activities that are part of copyrighted content.
• Sports fans often miss live games and rely on highlight reels. AI can automatically detect key moments like goals, touchdowns, or game-winning shots in uploaded UGC videos and compile them into personalized highlight reels.

How to use?

The information is returned with the video frame number where it was found and probability of the detected activty. Identification is done using AI, so for each activity a probability percentage is applied; activities with a probability of at least 30% are included in the response.

Video processing speed is approximately 1:5.

Example of detected sports activity:

{
  "sport_detected": true,
  "detection_results": [ "shooting goal (soccer)" ],
  "frames": [
      {
          "label": "shooting goal (soccer)",
          "frame_number": 98,
          "confidence": 0.99
      },...
  ]
}

Example response when sports activities were not found:

{
  "sport_detected": false,
  "detection_results": []
  "frames": []
}

Please note that the API only provides a set of data (json) about the objects found, so no video is generated. The demo video video (above ^) was specially created based on json from the API for visual demonstration and better perception of the possibilities.