Category: F2024 – Mang

Computer Vision for Artists and Designers provides a comprehensive overview of the technology, its history, and its potential applications in the creative field. While the essay highlights the advancements in computer vision, it also acknowledges the limitations and challenges that still exist.

One of the key points emphasized in the essay is the difference between human and computer vision. Humans possess a unique ability to interpret visual information based on context, prior knowledge, and emotional cues. Computers, on the other hand, rely on algorithms and data to process and understand visual data. While computers can excel at specific tasks like object detection and tracking, they often struggle with more nuanced aspects of human perception.

The essay also discusses various techniques used in computer vision, such as object detection, object tracking, feature matching, and deep learning. These techniques have enabled significant progress in fields like autonomous vehicles, medical imaging, and surveillance. However, it is important to note that these techniques are still evolving and may have limitations in certain scenarios.

The intersection of computer vision and art raises interesting questions about the role of technology in creative expression. While computer vision can provide new tools and possibilities for artists, it is also important to consider the ethical implications of surveillance and data collection. Artists should be mindful of privacy concerns and strive to use technology in ways that enhance rather than exploit human creativity.

Concept

For my midterm project, I decided to design an interactive photo lab experience. The user starts the program off with the choice of being a customer for a day or an employee and then that decision guides their experience. As an employee, they have more of a ‘mini game’ experience (comparable to Papa’s Pizzeria on CoolMath) while if they choose to be a customer, it’s more of a learning experience for a beginner film photographer (inspired a bit by Duolingo). My main motivation for this design was based on personal experience because I started working at a photo lab roughly 2 months after I shot my first roll of film so it felt like I was learning a ton as a photographer and an employee super quickly.  I also think it would be cool if my job back home could use this as a fun resource for first-time customers.

Code progress

So far, in terms of ‘code-progress,’ I am pretty satisfied with how things are developing. Because the program is split into two sections (customer and employee experience) I devoted this weekend to just working on the employee experience. Working out small kinks has been quite time consuming because of the complexity to with which I want the program to function but I believe it is still manageable (for now). For the most part, I’ve relied on Canva for basic ‘set-design’ just to lighten the load a bit, and for aesthetic reasons so I’ve been working in layers quite a bit.

For example, this is the base layer of the processing room in the employee section. I’ve added different ‘layers’ on top so that depending on the mouse location, the program produces different signifiers that prompt you to open one of the four mini games or return back to the home page.

Most frightening part

Overall, I don’t think the implementation of my ideas will be all that difficult compelte. However, I am frightened by how tediously time consuming it will be, because it’s already taken be a significant amount of time to being making headway on 2 of the mini games. Furthermore, I still haven’t solidified what I want the customer’s side to look like, so I’m afraid that it won’t balance out appropriately with what I’ve already created.

Lastly, there are smaller things I’m holding off on debugging until the framework is fully complete and I’m really hoping that doesn’t prove to be a huge mistake. I’ve implemented the windowWidth components to try and keep my program (mostly) adaptable to any window size but quickly that it ends up distorting some of the images in a weird way and and making the space a bit awkward at times. So, I’m hoping when I circle back at the end it is a fairly easy solution.

As you can see I am relying heavily on windowWidth and windowHeight for a lot of stuff and don’t really know what to expect/how to solve the issue if it starts to not perform the way I’d like.

Nevertheless, I will try to address these concerns by working in segments, as I have been doing. With a project like this, it’d be very easy to become overwhelmed by the ideas, and implementation and everything else going on so I’m trying to keep everything broken down, into more digestible mini projects for me to complete. More specifically, in regards to the formatting concerns I think I’m gonna start recording the ratios of the important objects just have an easily accessible reference for what I’m most concerned about (i.e. making sure the instruction manual is centered). If something small is a little off, or has some slight variation from time to time, I’ll to address it but don’t want to become too fixated on little things.

intro

For my midterm project, although the Object Life Sim project seems promising to me, I do not have much idea for furthering it, and I’d rather incorporate a higher degree of interactivity in this project. Therefore, I would like to integrate the elements that appeared in my previous attempts – both conceptually and technologically – into a new game. And for this progress report, I will start with composing a standard game design documentation and then touch on the technical difficulties and that of the desgin-wise.

game design documentation

1. Game Overview

Title: Motion Ships
Concept:
Motion Ship will be a space-themed action game where players control a spaceship using head movements detected via a webcam. Navigate through space environments, combat obstacles, and achieve high scores through maneuvering and sound-activated attacks.

2. Game Structure

2.1. Game Flow

1.  Start Screen
2. Configuration Menu
3. Gameplay
4. End Screen
5. LoopBack or Exit

3. User Interface

3.1. Start Screen

Title Display: “Motion Ships”
Introduction: “Control the spaceship with your head.”
Prompt: “Press any key to start.”
Functionality: Waits for any key press to transition to the Configuration Menu.

3.2. Configuration Menu

Customization Options:
    Spaceship Selection: Choice between two spaceship models.
    Background Selection: Choice among four distinct battle backgrounds.

Controls:
    Selection Interface: Allows players to select their preferred spaceship and background before starting the game.
    Functionality: After configuration, initiates gameplay.

3.3. End Screen

Score Display: Shows the player’s final score.
Functionality: Offers players the choice to restart or modify settings post-game.

4. Gameplay Mechanics

4.1. Player Controls

Input Methods:
    Webcam-Based Head Movement Detection: Controls spaceship movement along the X and Y axes.
    Sound Input Detection: Monitors ambient sound levels to trigger beam launches.

Control Scheme:
    Movement: The spaceship moves along the X and Y axes based on the player’s head position detected by a face-detection machine learning model.
    Rotation: The spaceship rotates in real-time to mirror the direction of movement, simulating realistic aircraft maneuvers.

4.2. Player Character

Spaceship Model:
Formats Supported: 3D .obj files.
Positioning:
    Fixed Z-Axis: The spaceship remains on a fixed plane along the Z-axis.
    Dynamic X and Y Axes: Movement is controlled via head tracking.

Energy System:
Energy Meter: Displayed on the bottom-left of the screen, indicating available energy.
Energy Consumption: Firing laser beams depletes the energy meter.
Energy Recovery: Collecting energy ores replenishes the energy meter based on the ore’s size.

4.3. Movement Mechanics

Forward Motion:
    Illusion of Movement: The spaceship appears to move forward by keeping its Z-position fixed while other objects move towards the spaceship.
    Relative Speeds: All in-game objects move at a base speed relative to the spaceship’s “forward” movement. Objects capable of independent movement have their own speed vectors in addition to the base speed.

4.4. Combat Mechanics

Attacking:
    Weapon: Laser beams automatically launched based on ambient sound levels.

Trigger Mechanism:
    Sound Input Detection: Continuously monitors the player’s microphone input.
    Threshold-Based Activation: When the ambient sound level exceeds a predefined threshold, the spaceship automatically fires laser beams.

Laser Behavior:
    Trajectory: Beams travel along the negative Z-axis from the spaceship’s X and Y positions when launching.
    Energy Cost: Each launched beam consumes a small amount of the spaceship’s energy.

Feedback Indicators:
    Visual Cue: Brief animation or flashing effect indicating that a beam has been fired due to sound input.
    Energy Indicator: Real-time update of the energy meter showing depletion upon beam firing.

4.5. Obstacles and Enemies

4.5.1. Meteoroids

Appearance: Rock-textured spherical objects.
Behavior: Typically remain fixed relative to the background.
Size Variation: Random sizes within predefined limits.

Interactions:
    Destruction: Requires a number of laser hits proportional to size to be destroyed. Upon destruction, the meteoroid shakes briefly before disappearing.
    Collision: If a meteoroid collides with the player’s spaceship, both are destroyed, resulting in game over.

4.5.2. Energy Ores

Appearance: Crystal-textured spherical objects.
Behavior: Typically remain fixed relative to the background.
Size Variation: Random sizes within predefined limits.

Interactions:
    Destruction: Requires a number of laser hits proportional to size to be destroyed. Upon destruction, the energy ore shakes briefly before disappearing, and the player’s energy is replenished based on the ore’s size.
    Collision: If an energy ore collides with the player’s spaceship, both are destroyed, resulting in game over.

4.5.3. Enemy Ships

Appearance: Alternate spaceship models not selected by the player during configuration.
Behavior: Dynamic movement across all X, Y, and Z axes relative to the base speed. Change direction randomly every few seconds.
Collision Avoidance: Actively avoid colliding with other objects except the player’s spaceship.

Interactions:
    Destruction: Requires a number of laser hits to be destroyed. Upon destruction, the enemy ship shakes briefly before disappearing.
    Collision: If an enemy ship collides with the player’s spaceship, both are destroyed, resulting in game over.

4.6. Object Spawning and Lifecycle

Spawn Mechanics:
Spawn Rate: Objects spawn at regular intervals.
Spawn Probability:
    Meteoroids and Enemy Ships: Equal likelihood.
    Energy Ores: Lower likelihood compared to meteoroids and enemy ships.

Despawn Conditions:
Objects are removed from the game once their actual Z-position surpasses the camera’s Z-position by a specified distance, ensuring optimal performance and resource management.

5. Environment and Visual Design

5.1. Backgrounds

Selection: Players can choose from four distinct battle backgrounds during the configuration phase.
Parallax Effect: Backgrounds subtly move in response to the spaceship’s movement to create a dynamic and immersive living space effect.
Camera Perspective: Fixed camera position to maintain consistent player orientation.

5.2. Lighting

Position: Fixed behind the camera.
Effect: Illuminates all objects within the 3D space, ensuring visibility and depth perception.

5.3. Graphics

3D Models:
    Spaceships: .obj models for both player and enemy ships.
    Obstacles: Spherical models for meteoroids and energy ores, with varying textures based on type (grey for meteoroids, green for energy ores).

Animations:
Destruction Effects: Shaking animations upon object destruction before removal.
    Movement Animations: Smooth transitions and rotations for moving objects.

6. Technical Specifications

6.1. Technology Stack

Machine Learning Integration: Utilize a face-detection library (ml5.js) for webcam-based head movement tracking.
Sound Processing Library: Integrate a sound analysis library (p5.sound) to process and analyze ambient sound levels for beam activation.

6.2. Performance Considerations

Object Management: Efficient spawning and despawning of objects to minimize memory usage and ensure smooth gameplay.
Collision Detection: Algorithms for real-time detection between laser beams and obstacles/enemies, as well as between the player’s spaceship and other objects.

6.3. User Input Handling

Face Detection: Implement real-time face and head movement tracking for responsive spaceship control through webcam access.
Sound Threshold Detection: Continuously monitor ambient sound levels to trigger beam launches when thresholds are met through microphone access.

7. Additional Features and Considerations

7.1. Scoring System

Score Incrementation: Points awarded based on actions such as destroying obstacles, collecting energy ores, and surviving over time.
High Score Tracking: Persistent tracking and display of the highest score achieved across game sessions.

7.2. Audio

Background Music: Ambient space-themed music to enhance immersion.
Sound Effects:

1. Laser Firing: Audio cues when beams are launched.
2. Destruction: Sounds for destroying objects.
3. Collisions: Impact sounds for collisions leading to game over.
4. Game Over: Sound indicating the end of the game.

7.3. Accessibility

Instructions: Clear in-game instructions or tutorials to guide new players on using head movements and sound inputs for control.
Visual Indicators: HUD elements like energy meters and score displays designed for clarity and ease of reading.
Adjustable Sensitivities: Allow players to adjust head movement and sound input sensitivities to accommodate different playstyles and environments.

8. Implementation Notes

8.1. Sound-Based Beam Launching

Calibration: Provide an initial calibration phase where the game adjusts the sound input threshold based on the player’s environment to prevent unintended beam firing.
Feedback Mechanism: Visual and audio feedback when beams are launched due to sound input, helping players associate their actions with in-game events.

potential challenges

1. Real-Time Face Detection and Head Movement Tracking

Real-time video processing and face detection can be computationally intensive, leading to lag or reduced frame rates. SO, implementing throttling mechanisms to control the rate of face detection processing could be conducive to overall efficiency.

2. Sound Input Detection and Processing

Differentiating intentional sound inputs (e.g., biu-biu-biu voice command) from ambient noises to trigger beam launches without false positives may need amplitude-based or frequency-based thresholding. It might be helpful to establish a continuous record (RMS, LUFS, etc.) before gamplay to calculate a based noise level.

3. Collision Detection in 3D Space

3D collision detection would be more complex and resource-intensive than 2D that I used in previous projects, especially with multiple moving objects and interactions. The potential solution could include:

• Early Exit Strategies:
  Implement early exit conditions where collision checks are skipped if objects are too far apart to interact, conserving resources.

• Simplify Collision Shapes – Axis-Aligned Bounding Boxes (AABB):
  For certain objects, AABB can offer a balance between simplicity and collision detection accuracy.

4. Object Spawning and Lifecycle Management

Without proper management, the continuous spawning of objects can lead to memory leaks and degraded performance over time, while the spawn rate and probability are important to maintain a balanced difficulty without overwhelming players or underutilizing game mechanics.

Therefore, adjusting spawn rates and object probabilities dynamically based on the player’s score or progression can be used to maintain engaging and balanced gameplay.

demo version

With all that said, I developed an early demo version without fine-tuning the parameters and visual effects.

Fundamentally, the first two technical difficulties are tackled in this stage, while the rest is untouched. Personally speaking, as this project expanded in scale, I learned about the game design structure and built highly object-oriented programming with code as specialized as possible.

Exploring the Pioneering World of Videoplace and Computer Vision

One of the most intriguing things about technology before the 2010s is how companies like Videoplace managed to harness ‘Artificial Reality’ between 1969 and 1975. This was long before smartphones and personal computers became mainstream, making their achievements feel ahead of their time. In fact, Videoplace was a pioneer in creating a VR-like experience, and this early attempt at computer vision offers us a glimpse into how computers and humans “see” differently.

Computer Vision vs. Human Vision

While human vision is holistic and context-driven, computer vision is all about analyzing images pixel by pixel. For example, Myron Krueger’s Videoplace used algorithms like the brightest pixel detection to track users’ movements. This means the computer focused on changes in brightness and motion, rather than understanding the full scene as a human would. Despite this difference, computers excel at tracking minute details or movements that humans might miss, making them ideal for tasks requiring precision.

Techniques for Helping Computers “See”

The article discusses techniques like motion detection, which helps computers identify changes in pixel values to track movement. These early algorithms now form the foundation of modern visual effects software like Adobe After Effects and Nuke, allowing them to “see” and track objects with high accuracy. As someone interested in filmmaking, I’ve realized how powerful this technology is for enhancing video quality and adding visual effects, making it possible to create dynamic, interactive experiences.

The Impact on Interactive Art and Surveillance

Videoplace demonstrated how computer vision can transform interactive art, allowing viewers to influence digital experiences in real-time. This ability to track movements made art installations more immersive, blending the boundaries between the physical and digital world. However, this same technology also lends itself to surveillance, as seen with systems like the Suicide Box, which monitored human behavior. This dual nature of computer vision—both as an artistic tool and a means of surveillance—reveals its vast potential and challenges in shaping interactive media. This shows how interactive media is not just limited to entertainment but also social political issues within society.

Final Thoughts 

Reflecting on Krueger’s belief that human interaction is vital in developing computer systems, it’s clear that his vision holds true today. Despite their differences, computer vision and human vision can work together to create compelling, interactive experiences. Understanding how to leverage this technology allows us to not just create art but to redefine how we engage with the digital world. It’s fascinating to realize that the more we harness computer vision, the more limitless our creative potential becomes.

 

Sources Used

• http://www.flong.com/archive/texts/essays/essay_cvad/index.html (The article
• Grammarly and ChatGPT  for grammar and paragraph formatting.

One fundamental difference between human and computer is that computer cannot easily distinguish between objects. They only recognize sources as pixels and algorithms have to be implemented to distinguish a specific object. What impressed me most in the reading is Myron Krueger’s Videoplay. The system he built has a diverse enough reaction bank that users can play with. Based on the movement of the participants, the computer generates visual cues accordingly to that input. Based on the participants’ actions, the computer generates visual cues accordingly. I think the ability to respond flexibly gives the system an impression of intellectual thinking, rather than just following rigid algorithms. I also noticed how Krueger interpreted physical interaction in human-computer interaction. For example, squeezing a virtual character on the screen causes it to pop and disappear (4:12), making the interaction between human and computer feel more realistic and reducing the sense of untouchability of objects on the screen.

Other techniques that I find interesting to be used in interactive design are detecting presence and simple interactions described by Jonah Warren. Because the way computer is used in daily life is mostly rigid and only tailored to specific needs, tweaking computer’s reaction based on the aforementioned input can create an interesting art media. For example, the Cheese Installation by Christian Möller is a unique way of interpreting computer data. Normally, we would not perceive the intensity of a smile in specific percentage. However, the way the machine interprets how we smile and return it into visual cue create a creative material for artwork. As such, a simple footage of smiling can be turned into a scrutiny of someone’s emotions.

Computer vision is an extremely interesting concept, and emerging technologies significantly boost its development. Although the author mentioned that machines were not capable of doing ‘generalized’ visual input analysis and were heavily dependent on the specific set of programming techniques that would allow them to recognize only certain real-world scenes, I believe that soon (if not already), thanks to the rapid growth of AI, computers will be able to receive and process the information in a ‘general’ way without the need for the special setup in the software. Still, such a technology will probably not be widely applied at the beginning. In one of the Core classes that I recently took, we were talking a lot about Machine Learning, and one of the techniques that are used in that field with the purpose of computer vision enhancement is the Convolutional Neural Network, or CNN which learns by filtering the inputs of text, images and even sounds. Other techniques that were mentioned in the reading include brightness adjustment threshold, background subtraction, and frame differencing. All of them are described to be used to help a computer recognize the visual patterns and track changes in them, but I am sure that now there certainly exist some highly advanced technologies that, although probably based on the techniques described by the author, but still work much more efficiently and maybe incorporate a number of other methods simultaneously.

The development of computer vision is obviously a large step in our technological and scientific progress. It can be used for a lot of good things aimed at improving our lives. In healthcare, for example, it can be used to help doctors diagnose and treat patients or even assist them during surgeries. It is also widely used in self-driving vehicles, which can potentially improve road traveling conditions by making them much more regulated and safer. Computer vision can also be used for leisure activities and entertainment like games – I remember playing Xbox 360 with Kinect, a device that could track your motions and allow you to play some very cool games where you actually needed to jump and do some physical activities in real life, and they would affect the actions of your character in the game. As for Interactive Art, a great example of computer vision implementation is the installation that we recently saw in class, where you can catch the letters with your arms.

However, by implementing all these advanced technologies like motion and face recognition, we should always keep in mind the potential ethical issues that can arise as the line between interaction and privacy becomes more and more blurred. Not to mention, surveillance devices can also raise concerns among people as they become more and more ubiquitous and use high-level precision that is trained by analyzing hundreds of thousands of patterns of human facial expression, emotional behavior, etc. There are many questions that put in doubt the legitimacy of the development of such technologies as human rights can be at stake, but I guess such a discussion can last for a very long time. In my opinion, we just need to keep in mind that each of us can have different perspectives and attitudes toward computer vision, and this is completely normal.

Concept

I was influenced by the optical illusions that I encountered during my childhood. Consequently, I resolved to develop something similar. The code uses the way motion, shape, and color interact with each other to make a dynamic visual experience that looks like an optical illusion.

Optical illusions use the way our brains handle patterns, color changes, and movement to make us feel things like depth, movement, or distortion that aren’t really there.

The squares’s lineup here make it look like the image is spinning and pulsing in the opposite direction when the user “pauses” the rotating effect by pressing the “Space” button because the speed of the movement changes abruptly. The slowly changing colors and the many concentric squares’ borders also mess with how we think about depth, drawing our attention inward while the colors in the background change easily, throwing us off.

Sketch

Code I am proud of

“For()” loops are very important to this sketch because they let you make grids of circular squares over and over again. The ‘for()’ loops make placing each square easier by going through different places over and over again. Without them, it would be hard to do by hand. The design of these loops is both organized and complicated, which improves the illusion itself.

// Loop for centers of every possible square patterns
for (let x = -500 + 50; x < 500; x += 100) {
  for (let y = -500 + 50; y < 500; y += 100) {
    // Variable stroke color based on a sine wave
    changeCol = sin(t * 100);
    let newCol = map(changeCol, -1, 1, 0, 255);

    // Draw multiple concentric squares with decreasing sizes and random RGB values
    stroke(newCol);
    strokeWeight(3);
    fill(r, g, b, 60); // Random RGB with fixed alpha
    rect(x, y, 80, 80);

    stroke(newCol / 2);
    fill(r, g, b, 80); // RGB for smaller square
    rect(x, y, 60, 60);

    stroke(newCol / 4);
    fill(r, g, b, 120); // RGB for even smaller square
    rect(x, y, 40, 40);

    stroke(newCol / 6);
    fill(r, g, b, 140); // RGB with different alpha
    rect(x, y, 20, 20);

    stroke(newCol / 8);
    fill(r, g, b, 160); // RGB with a different alpha
    rect(x, y, 10, 10);
  }
}

These loops also make the sketch scalable and easy to edit in the future.

Future Improvements

• There should be a live control panel with settings for things like spinning speed, background color transition speed, and square size. This would give people more control over the illusion and let them make the experience their own.
• Adding a z-axis or moving based on depth would make things even more complicated. The sense of 3D space could be better created by making the squares bigger in the center and smaller in the background to show depth.
• Random colors add variety, but a color pattern that works well together can make something look better.

This week’s reading was very interesting as it dives into the technicalities of computer vision, making it seem a lot more doable, than what one might think. By breaking down computer vision into its algorithm’s basic fundamental nature, it becomes something as simple as is this pixel’s brightness greater than or less than my threshold value. Although this type of algorithmic analysis is completely different from how we as humans use vision to understand the world, it does seem like an interesting idea to conceptualize. The farthest I can reach in comparison between computer vision and human vision (as someone with no visual impairments) might be in looking at a black-and-white photo or viewing something with a stark contrast filter.

These algorithms are super interesting though because we can literally just work from light comparison in order to send different messages to the computer about what it’s seeing. In order to optimize this it is important that we give the computer clear cut distinctions from what we want it to focus on and what we don’t. I also find it interesting how beyond light, we can use things such as heat to act as the eyes of the computer.

I think computer vision’s capacity for tracking and surveillance affects its use in interactive art because opens new avenues for creativity and expression through computers. I think with the term surveillance specifically, it can also significantly reduce an individual’s right to privacy for the sake of “art.” For example, the Suicide Box project, in my opinion is completely unethical and an abuse of ‘public privacy.’ However, that then stems the issue beyond interactive art because it becomes an issue of privacy and security. Nonetheless, because computer vision has a multitude of applications I don’t believe it is at all limited to what it can be used for in interactive art, which is why we need to stay on top of its usage to ensure people’s security.

Computer Vision has consistently been a prominent subject among programmers. It is a principal catalyst in the AI sector today, emerging as a viable option in the healthcare business and enhancing security. However, this reading not only provides a historical context to Computer Vision being the cold, technical niche used by military and mainstream industry; it has now become an accesible tool for artists, paving the way for an increased integration between computers and artists. This seems especially applicable in the modern day, when the lines separating engineering and art are becoming increasingly hazy and anyone can create interactive, visually stunning experiences.

The recurring theme in the reading is how artists have used Computer Vision to to build interactive exhibits that engage spectators in real time. Myron Kruger’s “VideoPlace” is a fitting example of this, turning a cold, impersonal technology to something deeply human-centered: using full-body interactions as a dynamic input to create immersive environments. In Computer Engineering, this is closely tied to the evolving user interfaces, which today powers technologies like motion sensors and gesture-based controls in gaming (Like VR).

Regarding the difference between Computer and Human Vision, one important difference between computers and humans is that computers use algorithms to understand images, but human vision is intuitive and contextual. The computer can “see” what “we” want it to by using techniques like frame differencing, background removal, and brightness thresholding, but these approaches are frequently inflexible, periodic and task-specific in contrast to the flexible nature of human vision. Another key difference is the number of input channels in human and computer vision. Humans take multiple inputs like colors, contrast, visual acuity and so on to make a cohesive perception, while computers only take a limited input based on the task they are intended to perform.

In interactive art, this rigidity in Computer Vision can actually be useful as it helps in simplifying the interaction to specific movements or gestures, allowing for a clearer, more focused experience for the viewers. However, as seen in the case of “Sorting Daemon” by David Robesky, such automated systems can profile people, leading to intrusive monitoring and raise other ethical concerns. As Computer Vision technology develops further, it is imperative to guarantee accountability, equity, and transparency in its application.

The topic of computer vision has been an increasingly popular idea, and I believe in the next couple years there will continue to be big leaps in its development and applications. To think that less than a decade ago computer vision was solely for the military and for those in higher education has now turn into a readily available technology for artists and the general public is amazing. As such, I really enjoy how the author presented the paper for beginners in computer vision and provided advice on how to approach projects with computer vision. The projects on tracking and surveillance were actually some of the projects that stuck with me the most throughout the reading. The Suicide Box project I felt on one hand sort of telling the story of the individual whom society, especially when their death wasn’t accounted for, but on the other hand super demoralizing because the project relies on the death of these individual. As such, I feel the capacity for tracking and surveillance for computer vision is a sensitive and difficult issue to uncover fully. There is of course a huge capacity and room for growth in the field of computer vision, however the ethnicity needs to be checked and balance with the morality and freedoms of individuals.

Through the reading, the author mentions how computer visions programs will need to be chosen mindful in order to optimally tackle the problem at hand. With complicated terms such as background subtract and frame differencing, I believe the complexity of differentiating between objects, backgrounds, and lighting is the biggest different from computer vision to human vision. As humans, we process millions of tiny information at once without noticing it and as programmers, the little things like recognizing an object are magnified as humans attempt to describe in code what that object is in computer language. Working with interactive media and computer vision for the rest of the semester, I believe the techniques regarding background subtraction and brightness tracking will play the biggest role in our projects. I feel many of our projects will rely on clear differentiating of human movements and interactions with the project. Without these techniques, I fear our projects may fall apart or not work as the screen would act as a camera and not a program which can be interacted with.