MorphCast Emotion AI – FAQ

MorphCast Emotion AI is a technology that enables the recognition and analysis of facial emotions (and not only) through artificial intelligence. It is a state-of-the-art technology that uses machine learning algorithms to analyze facial expressions and detect emotions in real-time. MorphCast Emotion AI is designed to be integrated into web applications, making it accessible to a wide range of users, including businesses and researchers. By analyzing and interpreting emotional data, MorphCast Emotion AI can provide valuable insights into customer behavior, market trends, and more. It is a powerful tool that can be used to improve customer engagement, optimize marketing campaigns, and enhance the overall user experience.

MorphCast fully embraces Rosalind Picard’s groundbreaking work in Affective Computing by creating technology that not only recognizes but also responds to human emotions. Our platform processes emotions and user states in real time, enabling applications in education, customer experience, and healthcare. Furthermore, we prioritize user privacy by processing all data locally within the browser, without relying on remote servers, ensuring a secure and trust-centered approach. MorphCast continues to advance Picard’s vision by developing tools that promote well-being, inclusivity, and a more empathetic interaction between humans and technology.

AI Agents are software programs designed to interact with their environment, process data, and make autonomous decisions to perform specific tasks. Unlike traditional software, AI Agents analyze context, learn from gathered data, and adapt their actions to achieve predefined goals.

By integrating technologies like facial emotion recognition, AI Agents can understand and respond to users’ emotions in real time. This enables more empathetic and personalized interactions, improving the overall user experience. (morphcast.com).

AI Agents are used across various industries, including:
Healthcare: Monitoring patient emotions to provide tailored support.
Customer service: Detecting frustration or confusion to adapt responses and improve customer satisfaction.
Education: Adjusting educational content based on students’ emotional reactions to enhance learning.
Digital marketing: Creating campaigns that adapt to user emotions to boost engagement.

MorphCast provides several solutions to create AI Agents, such as:
EmotionVoice GPT AI Agent: A cutting-edge platform for building empathetic voice agents without coding requirements. (morphcast.com)
Emotion AI HTML5 SDK: A lightweight SDK for integrating facial emotion recognition into web applications, empowering AI Agents with advanced capabilities. (morphcast.com).

MorphCast solutions process data directly within the user’s browser, without sending facial information to the cloud. This approach ensures users retain full control over their data, safeguarding their privacy. (morphcast.com).

For detailed information about what MorphCast Emotion AI is able to detect, please see this documentation about How our Emotion AI works.

The extracted data from our Emotion AI are meticulously structured to ensure ease of understanding and analysis. Users have the flexibility to download this data in widely recognized formats, including CSV and JSON, catering to diverse analytical needs. Additionally, our platform offers a comprehensive Dashboard, equipped with charts and graphs, for a visual representation of the emotional analysis. This multifaceted approach allows for an in-depth exploration of the detected emotional states, facilitating a seamless integration into your analytical frameworks. A complete documentation about interpreting data generated by our Emotion AI, you can consult this guide offering a clear, in-depth explanation of all the charts and analytical tools used in the MorphCast products.

MorphCast Emotion AI’s server-free processing technology offers privacy, real-time response, scalability, cost-effectiveness, and a low carbon footprint. For detailed benefits, visit: advantages of choosing our server-free Emotion AI.

The software can be used in several industries, including:
– Emotion analysis: MorphCast Emotion AI can be used to analyze and profile online audiences, collecting valuable data to enhance business strategies.
– eLearning experiences: The software can be used to create emotionally interactive lessons that adapt to individual learners’ pace and improve the effectiveness and fun of learning experiences.
– HR: MorphCast Emotion AI can be used to analyze candidates during recruiting phases, and to deliver adaptive training paths based on employees’ interactions, emotional response, and level of engagement to improve personnel’s soft skills and boost performance.
– Product innovation: MorphCast offers flexible and user-friendly tools for creating proof-of-concepts and innovative products, participating in calls, and finding funds to finance disruptive ideas.
– Digital advertising: The software can help to capture the attention of online users, improve audience engagement, and boost digital campaign results through personalized and impressive emotionally interactive experiences.

Our neural networks have been trained using custom datasets with huge amounts of faces, ranging from all ethnicities, ages and genders. Although the dataset is not shared, we acknowledge that there may be a possibility of race, gender, and age bias in the analysis results. However, we emphasize that the influence is not significant, and the MorphCast Emotion AI has been tested on standard test benches by an independent research center, confirming its accuracy among different ages, genders, and ethnicities.
For detailed information about how is the software trained and what about bias, please see this documentation about Responsibility, Transparency and Clarity.
We can confirm that the accuracy of our AI has been validated through tests on a standard market testbed, BU-4DFE, which included representation from various ethnicities such as Asian, Black, Hispanic/Latino, African American, and White. This ensures that our system is reliable and accurate in analyzing facial expressions across a broad demographic spectrum.
Anyway we invite you to take the possibility of bias into account and minimize its effects depending on the purpose of the analysis. It is essential to note that mitigating bias in AI algorithms is an ongoing challenge, and MorphCast’s approach to acknowledging its existence and working to minimize its effects is a step in the right direction.

The accuracy of MorphCast Emotion AI depends on various factors such as the lighting conditions and camera quality. However, our server-free algorithms have an accuracy level that is at par with the competitors who use server-side algorithms (see this independent publication: Emotion recognition in humans and machines using posed and spontaneous facial expression). The software is empowered by AI algorithms with minimal computation and size overhead, ensuring efficiency in resource utilization. Additionally, the large number of configurable parameters guarantees flexibility, while the default configuration has filters already tuned for common usages, making it ready-to-use. The intuitive interface makes it easy-to-use, and it is compatible with almost every modern browser. MorphCast Emotion AI has a modular architecture that ensures its scalability. It is capable of reaching even 20-30 analyses per second on PC devices and usually reaches seven analyses per second on most last-generation mobiles. The dynamic power-save optimizer is designed to optimize CPU and GPU load when available, making it a power-efficient solution.

Lighting Requirements: 300 to 500 lux is often recommended for comfortable visibility and clear recording indoors and 750 to 1,000 lux for outdoor environment. 

For optimal emotion detection, the recommended distance between the camera and the subject is typically a close-up range of about 30 to 60 cm (approximately 1 to 2 feet). This distance allows the software to capture facial expressions clearly. To aid in positioning, you can use an on-screen template to help users align their faces correctly. The size of the face template will adjust based on the distance, ensuring optimal positioning for accurate emotion analysis.

Ensure that the camera is positioned at nose height of the viewer and that the face is directed straight towards it.

In addition to brightness, ensure there are no direct light sources behind the viewer or strong lateral sources that could create shadows on the face.

Make sure users do not have obstructing elements, such as sunglasses, covering the face, as they can affect the accuracy of facial expression analysis.

If your camera supports it, you can configure the SDK with a higher input resolution to the Face Detector, allowing for the detection of faces even at distances greater than the indicated range.

Yes, the parameter ‘maxInputFrameSize’ in our SDK can be adjusted to accommodate different distances for facial analysis, such as for outdoor advertising (OOH) use. This parameter indirectly influences how far the face being analyzed can be from the camera. For detailed guidance on how to adjust this and other parameters, please refer to the technical documentation related to our SDK.

For measuring confidence, consider using the “quadrant” outputs from our MorphCast Emotion AI SDK, specifically the “Conductive” quadrant. This approach, broader than identifying a specific mood like “Confident,” aligns well with confidence-related states and is grounded in appraisal theory. It offers a more encompassing view of emotional states akin to confidence.

Due to ethical considerations, which we at Morphcast hold in high regard, our artificial intelligence engine, MorphCast, has been deliberately designed to not recognize faces and to not track the behavior of the same face across the sequence of frames in a video stream or camera feed. Originally, it was designed to track, in each frame, the largest face if there are multiple faces present. The subsequent frame does not retain any identity information from the previous frame. Consequently, if two faces in successive frames alternate in size relative to each other, the engine will track each one alternately, unaware that the faces belong to two different individuals.

The Attention Index is a dynamic measure that evaluates a user’s level of focus or engagement with the screen. It takes into account whether the user’s face is within the webcam’s field of view, head position, and other indicators of emotional and mood behavior. This index is continually calculated, providing real-time insights into the attention levels of users. Importantly, the Attention Index is scaled from 0 to 1, where 0 indicates minimal or no attention and 1 signifies maximum attention.
Positivity Index
The Positivity Index is derived from the Arousal and Valence dimensions, part of the 2D emotional space model. Arousal indicates the level of engagement, and Valence represents the emotional positivity or negativity. The Positivity Index is also scaled from 0 to 1, where 0 represents a completely negative emotional state and 1 represents a completely positive state. 
Adjustable Angle Parameter: In the context of the Positivity Index, we have an adjustable angle parameter. This parameter is key in interpreting the emotional state within the 2D space of Arousal and Valence. It helps project the emotional state onto a single dimension, simplifying the complex emotional state into a single positivity score. By adjusting this angle, you can customize how different emotional states are interpreted and represented as a single value.
Practical Application: This means that depending on the angle you set, the SDK interprets and combines the arousal and valence values differently. This feature allows you to tailor the Positivity Index to better suit the specific needs or context of your application, it offers a means to calibrate how various emotional states are quantified and understood, providing a nuanced approach to emotional analysis.

MorphCast is a U.S.-based company, but its facial expression analysis technology is subject to the Artificial Intelligence Act (AI Act) when used within the European Union or when impacting EU citizens. The AI Act, which came into force on August 1, 2024, adopts a risk-based approach, classifying AI systems based on their potential societal impact.
Notably, the AI Act prohibits the use of AI for emotion recognition in workplace and educational settings within the EU, unless intended for medical or security purposes. This means that any deployment of such technology involving European citizens in these contexts must comply with strict regulations.
To ensure compliance, MorphCast will follow the necessary procedures, including providing detailed information about its model training to the relevant European authorities when applicable. This reflects our commitment to legal and ethical AI deployment, even as a U.S. company operating internationally.
It’s important to note that while the AI Act became effective in August 2024, its provisions will be enforced gradually. For example, the regulations concerning high-risk AI systems will take effect on August 2, 2026.
MorphCast will continue to closely monitor regulatory developments and adapt to evolving requirements, ensuring the responsible and compliant use of its technology for European users and beyond.

Currently, there are no documented cases where MorphCast data has been used as legal evidence. This is in line with the current state of AI legislation, which has yet to define specific rules for the legal use of such technologies in judicial contexts.

Our training algorithms and specific data sources are currently subject to industrial secrecy.
The decision to keep these aspects confidential is driven by our duty to safeguard the substantial investments that have enabled the creation of our advanced AI technology. This approach is essential for maintaining competitive advantage and ensuring the ongoing viability and innovation of our products.
We understand the importance of transparency in the AI industry and are committed to upholding ethical standards and regulatory compliance. If in the future we decide to make our algorithms and data sources open source, we will do so in a manner that is public and transparent, while still respecting the interests of our stakeholders and the integrity of our intellectual assets.
We appreciate the interest in our methodologies and remain dedicated to advancing the field of AI in a responsible and sustainable manner, balancing the need for openness with the imperative to protect our investments and innovations.

In line with our policy on confidentiality and the protection of our intellectual assets, we do not disclose specific details about the composition and size of the dataset used for training MorphCast’s AI. Similar to how the exclusive recipes of renowned AI (and not only AI) products, remain secret, we maintain the confidentiality of specific aspects of our training process to protect our investments and competitive edge.
However, we can confirm that the accuracy of our AI has been validated through tests on a standard market testbed, BU-4DFE, which included representation from various ethnicities such as Asian, Black, Hispanic/Latino, African American, and White. This ensures that our system is reliable and accurate in analyzing facial expressions across a broad demographic spectrum.
For those who wish to conduct more specific and detailed analyses, we are willing to provide the necessary tools for them to personally test the efficacy of our SDK on a testbed of their choice. We strongly believe in the capability of our system to meet the most demanding requirements and are open to collaborations that allow our clients to personally verify its effectiveness and accuracy.

MorphCast is a registered trademark of MorphCast Inc., a Delaware Corporation founded in 2024 in California (USA) by Stefano Bargagni and the team working together in Italy on the same project since 2014, whose business was acquired by MorphCast Inc. Stefano is a serial internet entrepreneur with a background in computer science. Since 2014, the team has focused on researching and developing deep neural network algorithms for facial expression detection, receiving significant recognitions including four seals of excellence from the European Union’s Horizon2020 program and numerous mentions in Gartner research.
In terms of corporate social responsibility, MorphCast commits to operating with authenticity, transparency, integrity, and social responsibility. This includes a commitment to environmental protection, human rights, and personal data protection, as highlighted in their Code of Ethics and guidelines for the responsible use of Emotion AI.
The team’s collaborations with research institutes like the CVC Center in Barcelona and the MICC in Florence, and partnerships with international entities in remote learning and coaching, demonstrate its commitment to promoting sustainable and socially responsible AI innovation.
MorphCast’s future vision envisages a transformative era where emotion AI becomes a cornerstone of human interactions and between humans and robots. The goal is to develop an ecosystem where humans, avatars, and robots coexist, benefiting from the emotional intelligence provided by MorphCast’s technology. Check out this article written by Stefano; it can be very inspiring.

Yes, MorphCast’s Emotion AI technology, developed by MorphCast Inc., utilizes a diverse dataset that includes images and videos from people of different ages, genders, and ethnicities. This broad representation is crucial for training the algorithms to accurately analyze facial expressions across various demographic groups. The technology emphasizes impartiality and does not process biometric data for the unique identification of individuals, ensuring privacy and ethical use. For more information see Responsibility, Transparency and Clarity.

MorphCast’s Emotion AI technology is designed to estimate emotional states and apparent characteristics, such as age and gender, with varying degrees of probability and accuracy. The software, implemented as a JavaScript library, operates in the end user’s browser and does not store images or videos, ensuring data privacy. The accuracy of facial expression recognition can be influenced by factors like image quality, algorithm design, and the diversity of the training data. MorphCast’s algorithms, which employ deep learning techniques and are continuously improved through research, aim to minimize bias and provide reliable emotion recognition across different groups.

Yes, MorphCast’s AI technology is informed by the research of leading experts in the field, including the work of Paul Ekman. The AI models used in MorphCast digitally reproduce the results from longstanding research and papers in the domain of emotion recognition and analysis. By grounding our technology in well-established scientific studies, we ensure a robust and credible foundation for our emotion AI systems.
MorphCast’s AI is trained using a variety of scholarly sources that contribute to its accuracy and reliability. Some of the key research works that have influenced our AI models include:
Russell, J., Lewicka, M. & Niit, T. (1989). A cross-cultural study of a circumplex model of affect. Journal of personality and social psychology, 57, 848–856.
Scherer, Klaus. (2005). Scherer KR. What are emotions? And how can they be measured? Soc Sci Inf 44: 695-729. Social Science Information. 44. 695-792. 
Ahn, Junghyun & Gobron, Stéphane & Silvestre, Quentin & Thalmann, Daniel. (2010). Asymmetrical Facial Expressions based on an Advanced Interpretation of Two-dimensional Russell Emotional Model. 
Paltoglou, G., & Thelwall, M. (2012). Seeing stars of valence and arousal in blog posts. IEEE Transactions on Affective Computing, 4(1), 116-123. 
Paul Ekman Basic-Emotions.

As with any technology that involves the collection and processing of personal data, there are potential ethical concerns associated with using MorphCast Emotion AI. These concerns primarily revolve around issues related to privacy, fairness, and potential bias.
One major ethical concern is the potential misuse or mishandling of personal data, particularly emotional data, that may be collected through the use of Emotion AI. To address this concern, MorphCast Emotion AI has established a code of ethics and guidelines to ensure ethical and responsible behavior from all parties involved in the use of the technology. The guidelines emphasize the importance of obtaining explicit and informed consent from individuals before collecting or processing their emotional data, respecting the privacy and dignity of individuals, and being transparent about the capabilities and limitations of Emotion AI systems.
Another ethical concern is the potential for Emotion AI to perpetuate or exacerbate existing biases or discrimination. This could occur if the training data used to develop the algorithms for Emotion AI is biased or if the technology is used to make decisions that discriminate against certain groups of people. To address this concern, MorphCast Emotion AI has trained its neural networks using custom datasets with huge amounts of faces, ranging from all ethnicities, ages, and genders. Additionally, MorphCast Emotion AI advises its customers to take into account the existence of any bias case by case in order to cancel or minimize its effect as much as possible.
Finally, there is the concern that Emotion AI could be used to manipulate or exploit individuals. This could occur if the technology is used to intentionally or unintentionally manipulate an individual’s emotional state or to exploit their emotional vulnerabilities. To address this concern, MorphCast Emotion AI advocates for a responsible and ethical approach to the development and use of Emotion AI systems. Ethics and Guidelines for a Responsible Approach to Emotion AI emphasize the importance of designing and using Emotion AI systems in a way that respects the privacy and dignity of individuals and avoids using the technology in a way that discriminates against or unfairly disadvantages any particular group of individuals.
In summary, while there are potential ethical concerns associated with using MorphCast Emotion AI, the company has taken steps to address these concerns and promote a responsible and ethical approach to the development and use of the technology.

MorphCast’s SDK is designed to be GDPR-compliant and adhere to all privacy regulations, ensuring maximum protection of user data. The core principle behind it is that emotional data analysis happens locally within the user’s browser, without ever transferring any personal information to MorphCast’s servers.
This architecture provides key privacy advantages:
Local data processing: Emotional analysis is performed entirely within the user’s browser, with no images or personal data being sent to external servers. MorphCast does not collect, store, or transfer any information that could identify the user.
Anonymous data transmission: The only data sent to the servers are aggregated emotion metrics associated with the experience’s timestamps. No user-related data, such as IP addresses, cookies, or other identifying information, is ever transmitted.
Complete anonymization: The data transmitted is fully aggregated and anonymous, not pseudonymous. This means there is no way to trace back to the user’s identity, unlike competitors who often rely on cloud servers to analyze facial expressions and may process images of the user, which exposes them to privacy risks.
In summary, MorphCast’s approach ensures full compliance with privacy regulations, protecting users from exposure of personal data and providing a secure alternative to cloud-based solutions that require the transfer of sensitive information. More details on this topic you can find here: privacy is an advantage. Please make sure to review the Privacy FAQ as well.

Yes, MorphCast Emotion AI can be integrated with other systems. MorphCast Emotion AI offers a large number of configurable parameters, which allows users to tailor the software to their specific needs. The integration of MorphCast Emotion AI with other systems can enhance the capabilities of those systems, enabling them to detect and analyze human emotions, and provide a more personalized and engaging experience for the end-users.

MorphCast offers a wide range of products as a service, including video and photo analysis, integrations with chatGPT, avatars, video conferencing, and more. The company has also recently launched a web apps place featuring an increasing number of ready-to-use Emotion AI web applications.
Stay updated on all the services and products of MorphCast Emotion AI by consulting the menu item “Products” in our site.
The company also provides professional services such as consulting, training, and support for products and services.

MorphCast is known for being the most cost-effective facial Emotion AI service on the market.
Since we have no cloud AI computing costs to bear, we can offer our service at a price up to a thousand times lower than that of our competitors, meanwhile reducing the carbon footprint of our products.
We offer a dynamic pricing model that revolves around the usage time of license keys on a monthly basis. You start with a significant quota of free usage each month. If your usage surpasses this free limit, you’ll automatically move up to a higher tiered plan. Conversely, if your usage decreases in the following month, the plan will automatically adjust to a lower tier, reflecting your reduced usage.
On the MorphCast site we provide complete Plans and pricing.

Yes, if you are affiliated with a university or are conducting research, you may be eligible to obtain a no-cost license for non-commercial use. Startups may also receive special consideration for such licenses. Please proceed to acquire the necessary license through our web site channel, and then send us an email (info@morphcast.com) to inform us of your status. This will allow us to verify that you qualify for a no-cost, non-commercial license.