Of the record-breaking $500 million 5-year NBA exclusive partnership in China, Tencent has only used 20% of the NBA content it has the rights to broadcast online. This translates into $400 million of commercial wastage: missed opportunities to build online viewership through broadcasting never-seen before clips of basketball legends.

With only 2 years of the partnership remaining, Tencent’s business objective was to efficiently maximize the value of its NBA content to add more paid viewers to its existing 40 million subscribers. To attract more subscribers, Tencent’s brand objective was to showcase its ability to provide more engaging video experiences for NBA fans as compared to traditional TV broadcasters by creatively using Tencent’s NBA video content.

Tencent could not achieve its goals alone. Enter IBM AI Vision: an AI-powered video editing solution enabling Tencent to realize both its business and brand objectives.

The crowded scenes, multiple camera angles and visual complexity of NBA players’ fast movements make basketball video analysis among one of computer vision’s most difficult tasks.

With over 170 million viewers in China, the NBA Finals is as big as it gets. During the most anticipated time of the NBA season –IBM would use AI technology to produce a video of the season’s FMVP within seconds after the MVP is announced. Compared to human video editing which can take days, IBM only needed 24-seconds to create a video clip of a Kevin Durant’s entire career highlights up to the MVP announcement. The video would maximize viewer engagement during the finale of the basketball season with just a simple click of a button which ultimately showcases IBM’s technology capabilities.

From facial recognition to jersey number recognition to player action detection, IBM AI Vision can sift through millions of video data to interpret, cut and create a professional quality video clip. This advanced AI technology can detect the tiniest image details such as a player’s facial happy or angry expressions ranging and basketball actions such as dunks or blocks. In other words, within seconds, millions of video data files pass through IBM’s special algorithm designed to produce a 24-second video clip showcasing any NBA player’s unique career highlights and signature moves.

The marketing campaign objective was to produce a highly engaging online video to showcase IBM’s technology capability. What better way to achieve this than an AI-powered video clip produced in real-time at the height of playoff finale excitement? By creatively, cleverly and efficiently using the massive database of NBA video data, we successfully created the most relevant content at the right moment to engage 143 million highly engaged Chinese NBA fans.

Thus to capture the right clips, IBM worked closely with Tencent weeks before the 2018 NBA playoff season to create a special algorithm for its AI video technology for the final NBA game. The algorithm was designed to first interpret an NBA player via facial recognition and jersey number, by detect the various facial expressions and basketball actions such as dunks and passes of each player. The basketball action was then checked for success from ball and net tracking. Based on preset parameters for each player, IBM could then define successful actions, signature moves and unique emotions of top NBA players to create a video clip.

The final application of IBM’s data-driven AI Vision technology automatically produced captivating professional-grade video content within seconds instead of days. By creatively using video data and broadcasting it in real-time at the peak of the NBA playoff final game.

Consumer Experience & Behaviour Metrics

- 622 Million Total IBM AI Brand Exposure

- 3.2 Million Additional Audience Engagement

- 6.6 Million Earned Media Exposure

In total $3.5+ Million Incremental Earned Media Value.