Emotion Graph

Our Emotion Graph is the key feature of our Video Emotion Recognition Tool, providing AI-powered visual insights into emotions detected in video content. Built on the Emotion Wheel theory, the Emotion Graph illustrates primary emotions like happiness, sadness, anger, fear, surprise, disgust, contempt, and neutrality. We have also integrated the Valence-Arousal model to enhance real-time emotional analysis during video playback.

We offer two ways to visualize emotional insights with the Emotion Graph:

After the video processing is complete, the first view on the dashboard is the Static Emotion Graph. This graph presents a numerical breakdown of eight emotions, proportionally displayed with the sum of emotional values equaling 100%. It provides an overview of how each emotion is distributed across all visible faces in the video (you can also view actor by actor or phase by phase by selecting any face to see that individual's emotion analysis). The static graph is a simplified circle chart, showing the percentage of each detected emotion but without valence or arousal metrics, which are only available in the dynamic view.

Once the user starts playing the video, the Static Emotion Graph transforms into the Dynamic Emotion Graph. In this mode, a moving dot provides real-time, frame-by-frame emotional insights, powered by the Valence-Arousal model. The graph is displayed as a circular wheel, divided into four quadrants, showing:

Valence: How positive, neutral, or negative the emotions are.

Arousal: Whether the emotions are exciting or calming.

For example, during highly positive and exciting moments, the dot moves toward surprise in the high arousal, positive valence quadrant. If anger appears during a high-energy scene, the dot moves toward the negative valence quadrant. This dynamic view offers granular insights, enabling users to track emotional shifts throughout the video.

Pause and Analyze:

Even when the video is paused, the Dynamic Emotion Graph remains visible, providing an instant snapshot of the emotional state at that moment. Below the graph, you’ll see detailed numerical values for valence, arousal, and intensity, ranging from -1 to 1, giving you precise insight into the emotional strength and balance at any given point.

By combining the Emotion Wheel theory with the Valence-Arousal model, our Emotion Graph offers a powerful way to explore and analyze emotional responses in video content, making it a valuable tool for content creators, researchers, and marketers.