v2

As a Technical Product Manager, I led the development of GENIE, a state-of-the-art tool for feature comparison in images and age rating classification. In a hybrid software and management role, I scoped business use cases, delivered technical solutions, and successfully deployed GENIE to production.

Project Highlights

  • Objective: Enhance artwork personalization and age classification across platforms such as Netflix, Amazon, iTunes, and Spotify.

  • Core Functionality: GENIE utilizes advanced computer vision and machine learning (ML) techniques to analyze and compare feature similarities in images. It incorporates natural language processing (NLP) for metadata annotations and verifies age ratings based on content analysis.

  • Impact: Delivered personalized and culturally meaningful artwork recommendations that improve user engagement and discovery.

Key Insights: The Power of Visuals

Research revealed that cover artwork is the primary driver of content engagement, capturing 82% of user focus within 1.8 seconds.

  • GENIE optimized this by leveraging emotion recognition and complex facial expressions in artwork to convey nuanced tones, shown through A/B testing to drive higher engagement.

Fig. 1: Engagement of Different Cover Artwork

Fig. 1: Engagement of Different Cover Artwork

Generating Recommendations

GENIE leverages the importance of regional and cultural nuances in artwork personalization, as highlighted by Netflix’s success with titles like Sense8. While great stories resonate globally, tailored imagery helps enhance discovery by aligning with regional preferences.

Product Scope and Use Cases

GENIE’s scope spans platforms like Netflix, Amazon, YouTube, Hulu, Spotify, and HBO, focusing on:

  • Personalized artwork recommendations.

  • Detecting and removing near-duplicate images to streamline movie recommendations.

By addressing regional differences and optimizing visual content, GENIE ensures greater engagement and user satisfaction across diverse markets.

Fig. 2: Example of Netflix Image Feature Similarity Algorithm

Fig. 2: Example of Netflix Image Feature Similarity Algorithm

Fig. 3: Artwork Compliance Use Case

Fig. 3: Artwork Compliance Use Case

  1. Feature Matching: Utilized SIFT in OpenCV for structural similarity and histogram-based distance analysis.

  2. Robustness: Techniques are agnostic to borders, text, watermarks, and size/resolution variations, with adjustable thresholds and ratios.

  3. Detection Capabilities:

    • Borders: Border detection for image processing.

    • Text: OCR text detection using Tesseract OCR.

    • Watermarks: Watermark detection for enhanced image analysis.

  4. These techniques enable precise and adaptable image analysis across diverse use cases.

Fig. 4: Genie’s Feature Matching Demo Output

Fig. 4: Genie’s Feature Matching Demo Output

Project Impact

  1. Artwork Generation for Episodic Content
    GENIE uses video frames to create personalized artwork recommendations for episodic content across multiple platforms.

  2. Cultural DNA Extraction
    GENIE identifies emotional and classifiable elements from images or video frames, correlating them with annotations to generate culturally and locally resonant recommendations. Filters such as age ratings and extracted features enhance the precision of these recommendations.

  3. A/B Testing for Artwork
    GENIE generates artwork recommendations for cover art, allowing studios to create and review designs for A/B testing. Salient features are detected via object recognition to suggest optimized visuals, such as regional actor highlights, localized text, or borders.

  4. Age Rating Verification and Explicit Content Filtering
    GENIE uses computer vision and subtitles to verify age ratings and filter explicit content (e.g., nudity, violence, unauthorized logos). The tool provides predictive feedback for improved content annotation and ranking accuracy.

These use cases demonstrate GENIE’s versatility in enhancing personalization, cultural relevance, and content safety across platforms.

GENIE transformed the streaming media landscape by elevating artwork personalization, ensuring image compliance, and driving user engagement through tailored visuals. Its advanced algorithms empowered platforms like Netflix to create culturally resonant, emotionally compelling artwork from video frames, enhancing content discovery and user satisfaction. By enforcing compliance with age ratings and filtering explicit content, GENIE maintained content integrity while adhering to global standards. This groundbreaking tool set a new benchmark for personalized media experiences and compliance, leaving a lasting impact on the streaming industry.