With the popularity of short-form video content, maintaining audience interest through sports highlights is key for virality. This means creators and sports broadcasters have an incentive to post only the most thrilling and high impact highlights, like goals, to capture engagement. However, manually scrubbing through hours of live game footage to capture the timestamps of these fast-paced plays is inefficient and unscalable for real-time social media posting. The Find Events task on the Abilities tab can act as a tool, determining if a video contains a specific highlight event and locating those occurrences throughout the timeline.

For example, with soccer game footage, a segment of video should be flagged with the label highlight if it clearly shows the continuous sequence of a goal being scored.

We will need to separate normal midfield gameplay, throw-ins, and post-goal player celebrations from the exact active moments the scoring play actually occurs on screen.

Our expected inputs are videos, and the expected output will be the timestamps identifying exactly when the highlight sequence appears throughout the footage.

‍

UI Tutorial

Step 1: Create an Ability

Go to the Abilities tab and select the button Create Ability. Get early access to Abilities here >

Fill out basic information about the ability such as its name and the description of the task itself. Since we are classifying events in a video, select the Task Type as Find Events.

‍

Step 2: Task Configuration

To configure the task, we need to select a dataset for the specific task. If you have already uploaded your videos in a dataset simply select the name of your dataset. However, if you haven’t already done so then select <New Dataset> and upload your videos, label them by identifying when the goal appears in the broadcast, and create a label highlight in Event Names. 

Step 3: Configuration

Our next step is to configure the prompt, select the model, and image size. For this use case, we recommend using the below prompt and settings for highest accuracy and best results. Replace the soccer with the sport you would like to capture and add information about what defines a highlight.  

‍

Prompt: 

Analyze the provided video footage to determine if it depicts a soccer highlight, specifically a goal being scored.

Read the following definitions carefully before making your decision:

highlight: This scene clearly shows a goal being scored. Visual indicators include... Get early access to Abilities here >

Step 4: Run Evaluation

To check how well the prompt does against the dataset, our next step is to run the evaluation. If needed, review the examples in your dataset to ensure all necessary images can be used in the evaluation. 

‍

‍

Step 5: Check Evaluation

All evaluations can be reviewed in the Abilities tab by clicking the dropdown arrow next to the associated ability-alias. Evaluations can take around 15-20 minutes to complete based on the size of the dataset.

In addition to the performance, recall, and precision percentages on the abilities tab, you can see a visualization of what the model predicted by revisiting the dataset. Click on the three dots and select “Go to reference dataset”. 

Select one of the videos in the dataset and click on the review button. 

After running the evaluation you can see what the model labelled as highlight and compare it to what you labelled. With this, you can improve your prompts and thus improve your accuracy. 

‍

Tips for Accuracy

1. Define "Edge Cases"  The key to high accuracy is a deep understanding of your specific acceptance criteria. In a soccer context, the line between "highlight" and "other" can be thin. You must be explicitly clear about where that line is drawn.

2. Updating FPS Depending on the size and location of the logo you are trying to find, update the FPS accordingly. For smaller logos use a smaller FPS (3-5) and for larger logos that appear for longer use 6-10.

‍