1. Three Applications of Computer Vision on Your Phone
- Face Recognition: Unlocking your phone or authenticating apps.
- Text Recognition: Identifying and extracting text from images, such as translating signs or scanning documents.
- Augmented Reality (AR) Filters: Snapchat, Instagram, and other apps overlay graphics onto real-world scenes.
Source: OpenCV
2. Take a Photo of Your Desk and Ask ChatGPT to Identify Objects
Steps Involved in the Identification Process:
- Image Capture: Take a photo of the desk.
- Preprocessing: Adjust image size, brightness, or remove noise.
- Feature Extraction: Detect shapes, edges, textures, or colors.
- Object Detection and Classification: AI models identify and label each object.
- Output: Recognized objects are listed, sometimes with confidence scores or positions.
Challenges for ChatGPT in Object Recognition:
- Ambiguity: Similar-looking objects may be misidentified.
- Partial Visibility: Hidden or overlapping objects may not be detected.
- Context Understanding: ChatGPT identifies items but does not understand their purpose or relationship.
- Image Quality Dependence: Low-resolution or blurry images reduce accuracy.
Highlighting or Circling a Specific Item:
- ChatGPT cannot directly mark images but can suggest methods using code (e.g., Python/OpenCV).
- Misidentification or incorrect coordinates are common challenges.
Takeaway: Object recognition combined with image manipulation requires specialized tools beyond text-based AI.
3. Applications of Computer Vision That Are Not Trustworthy or Safe
- Surveillance/Facial Recognition: Using personal images without consent can violate privacy. Algorithms may reinforce biases, leading to discrimination.
- Autonomous Weapons: AI errors could cause accidents or harm, highlighting legal and ethical obligations for developers.
- Financial Verification Systems: Biometric errors could lock out legitimate users or allow fraud.
- General Ethical Risks: Developers should ensure transparency, communicate data collection methods, and adopt privacy-preserving practices (like homomorphic encryption).
- Collective Responsibility: Organizations, developers, and stakeholders must collaborate to create ethical frameworks and foster public trust.
Source:XenonStack