Vision

Prof. M. Salameh / Vision

Name three applications of Computer Vision you can recognize on your phone

Common applications include:

Face recognition for unlocking the phone
Object detection in camera apps (e.g., identifying pets, food, or landmarks)
Augmented reality filters in social media apps like Instagram or Snapchat

Take a photo of your desk and ask ChatGPT to identify or name all the objects

ChatGPT can analyze the photo and recognize objects such as a laptop, notebook, pen, phone, and cup. It uses computer vision models trained on large image datasets to label items accurately.

How many steps are involved in this identification process?

The process typically involves:

Image preprocessing (resizing, normalization)
Feature extraction using a neural network (e.g., CNN or Vision Transformer)
Object detection or classification
Label assignment and confidence estimation

In total, around 3–4 main steps.

Can you identify any challenges for ChatGPT?

Yes, there are challenges such as:

Low lighting or cluttered environments that make objects harder to detect
Ambiguous or partially hidden items
Dependence on image quality and camera angle

Ask ChatGPT to highlight or circle a specific item on your desk. Describe the results and possible challenges.

ChatGPT can successfully mark the requested item, such as circling a notebook or a mug. However, challenges may occur when objects overlap, lighting causes reflections, or the item looks similar to others nearby.

Several applications involve decision-making processes that rely on Computer Vision. Name some applications that you believe are not trustworthy or safe.

Potentially unsafe or untrustworthy uses include:

Facial recognition in public surveillance due to privacy and bias issues
Autonomous weapon systems using computer vision for targeting
Emotion detection systems used for hiring or policing decisions

These raise ethical, safety, and fairness concerns.