WorstPlans.com updates every Monday!

Your weekly source for terrible plans and ideas!

Category: Cameras

Stop getting hit by self-driving cars with this one fashion trick that involves putting weird labels on all your clothing! Don’t be the last one to catch on to this new fashion trend.

Background:

In a hypothetical future where self-driving cars are increasingly common, they’ll have to do a really good job of automatically distinguishing between things that require sudden braking (e.g. a person in the roadway) and things that are OK to hit (e.g. a tumbling empty cardboard box).

The issue:

This is a hard problem. When a car gets data from its various cameras (and other sensors), it needs to figure out what exactly it is that it is seeing (Figure 1).

1-self-driving-example.png

Fig. 1: This is probably a pedestrian in the roadway, but could it also be a billboard advertisement hundreds of feet away?

Although the specific “distant-billboard-or-close-pedestrian” question in Figure 1 can be answered just by using two cameras to estimate distance, there are situations where the problem must be resolved in a more complex fashion (Figure 2).

2-self-driving-problem.png

Fig. 2: Top: the image is interpreted correctly, and the car does NOT hit the pedestrian. Bottom: the car incorrectly believes that it sees a sunflower, and collides with it at full speed. Lest you think this is totally implausible, check out some specially-crafted adversarial examples (that can turn a panda into a banana) and a method of tricking lane-following algorithms into swerving the car into oncoming traffic.

Proposal:

We propose to place special “this is a human” symbols on articles of clothing that a human might wear (Figure 3).

When a car sees one of these unusual QR-code-like symbols, it will instantly say “ah, sunflowers do not wear specially-marked shoes, time to hit the brakes!”

To avoid this becoming a fashion disaster, these markings would not be apparently at normal human-visible wavelengths of light, but would only be detectable by special camera equipment.

Perhaps the markings could have fluorescent ink in them, and all cars could drive around with UV lights in the front.

3-shoes-with-markings.png

Fig. 3: Left: this is what the shoe looks like to a human—the markings are invisible to the naked eye. Middle: the camera can see wavelengths of light beyond human ability, and can detect these special markings (shown here as yellow checkerboards). Right: the camera sees the checkerboard, and the object-classification algorithm realizes that this shoe is likely to be attached to a human.

One common objection to many self-driving-car-related issues is “couldn’t some criminal put these markers all over the city, to trick self-driving cars?”

The answer is yes, but it would be as equally illegal as it currently is to put mannequins on a winding road (which would also confuse human drivers).

Conclusion:

This might be redundant with an infrared camera—in most locations, a human already is obviously distinguished from the background environment just by their warm-blooded glow in the infrared spectrum.

PROS: This will definitely make me a ton of money when it is licensed by major car manufacturers. Also, would someone please apply for and pay for a patent on my behalf? Thanks!

CONS: If one of these specially-marked shoes falls onto the roadway (perhaps by falling out of someone’s messenger bag while they’re biking), do we really want every car to come to a screeching halt at the sight of a single unattached shoe?

 

 

Improve the odds of finding a lost pet with this over-engineered license-plate-based system! The ultimate computer vision project for a machine learning startup.

Background:

“Lost cat” and “lost dog” signs are often placed up on telephone poles (Fig. 1), but it’s unlikely that a specific person who sees a lost pet will also have seen the sign (or even know that the pet is actually lost in the first place).

 

1-lost-snake.png

Fig. 1: A person who sees this sign will know to be on the lookout for a lost snake, but the chances of seeing both the snake AND the poster are quite low.

Proposal:

In order to add more people to the lost-pet-searching process, the proposed system is as follows:

On the searchers side:

  • Car owners can add a camera to their car (see license plate example in Figure 2) that constantly scans for unidentified animals. This requires no effort on the part of the driver.
  • The camera saves snapshots and GPS coordinates for every animal it sees, and uploads these to a “Find a Lost Pet” web site. Many of these animals are probably not lost, or even pets!

On the pet-recoverers side:

  • Anyone with a lost pet can post the details of their lost animal and a reward to the “Find a Lost Pet” site. Ideal information would include a photo, approximate location, and the owner’s contact information.

Once the “Find a Lost Pet” image analysis system detects a match between an uploaded image and a lost pet, a “bounty” is issued for the recovery of that pet, and nearby drivers are notified.

Finally, assuming the animal is safely returned in the same number of pieces that it was expected to be in (generally this number is “one”), the bounty is split three ways: the web site, camera owner, and animal-recoverer all get a fraction of the total reward. This aligns everyone’s incentives and encourages people to install pet-scanning cameras in the hope of a payout.

2-license-plate.png

Fig. 2: This license plate camera is a “dog-scanner” camera that is constantly on the lookout for unidentified potentially-lost animals. Backup cameras like this already exist, so producing the hardware for this system would be relatively straightforward.

PROS: This system will help find lost pets, and definitely won’t be repurposed to create a totalitarian police state.

CONS: Not especially useful in finding burrowed or aquatic animals, so try not to lose one of those.

Become a prey animal by putting eyes on the side of your head. Makes you a safer driver, but also encourages packs of howling wolves to attack you, so beware!

Background:

Human vision is limited to a ~180° horizontal angle (Figure 1) and an even smaller vertical angle.

This means you can easily be blindsided by objects coming from behind you. In ancient times, this was less of a worry, but with electric cars, electric scooters, and bicycles, there are a huge number of fast-moving and potentially-lethal objects that humans must be aware of.

 

1-normal-vision-cone.png

Fig. 1: Normal human vision only uses less than half of the total potentially-available visual information. Many prey animals have nearly 360º vision, so clearly this limitation is not an inherent limitation of biology.

Proposal:

Since cars are now the “apex predator” that is ranked above humans in the food chain, humans should adapt and become prey animals. This requires a visual adaptation to allow for 360º vision, which can be accomplished as shown in Figure 2.

 

2-360-degree-vision-lenses.png

Fig. 2: These “earmuff cameras” feed into the glasses in front, which provide a highly-warped fisheye view of the wearer’s environment. Although binocular vision may be impaired, the benefits of total visual awareness cannot be overstated.

Conclusion:

The example in figure 2 requires complicated electronics, but there’s probably a way to create an optical 360º-vision system that uses no electronics.

A similar product already exists: rear view mirror spy glasses—inexpensive sunglasses with mirrored “wings” that allow you to see behind you.

PROS: May reduce the number of deaths and injuries from accidents caused by a lack of visual situational awareness.

CONS: Can the human visual cortex handle this type of input data? Only one way to find out—experiment on some undergraduates.

Finally, you can become an ant, thanks to the power of VIRTUAL REALITY.

Background:

Currently, there is no easy way to have the experience of becoming a tiny ant [*]. This is a shortcoming that could not be addressed—until now, thanks to modern VR technology!

[*] You could watch the 1989 film Honey I Shrunk the Kids, but that isn’t an interactive experience.

Proposal:

Thanks to virtual reality, you can become an ant in 3 steps:

  1. Get a VR headset.
  2. Create a small remote-controlled car with two cameras on the front.
  3. Set up the R.C. car cameras to transmit to the VR headset.

Figure 1 shows the result of these steps.

Now you can be an ant!

1-vr-ant

Fig. 1: Left: someone wearing a VR headset that receives a pair of video signals from the remote-controlled car (orange) shown in the magnifying-glass inset (right)

 

2-vr-be-an-ant

Fig. 2: The experience of the viewer in VR goggles is shown at right. This is definitely exactly what an ant looks like close-up, as anyone who has seen the “Planet Earth” series can confirm. That’s how you know that this image was drawn with extensive consultation of reference material.

Conclusion:

This “ant VR” system theoretically be used for other purposes as well; maybe the ant-sized drone could check for cracks in hard-to-access parts of bridges or buildings, or an aquatic version could swim through a city’s water system to allow maintenance personnel to both look for leaks AND ALSO pretend to be an eel at the same time. Finally!

PROS: Lets you feel kinship with your insectoid brethren, the ants.

CONS: After spending a while in VR, you might think you actually ARE an ant and become unable to participate in human society.

3-vr-ant-original

Figure 3 (bonus): An extremely detailed technical schematic that will be used for manufacturing.

 

Solve your conference call woes with this one insane tip! Never lean your head weirdly in front of a laptop camera again. FINALLY.

The issue:

During a conference call, it can be difficult to position multiple people in such a way that everyone is actually in-frame.

Usually, either:

  1. Only one person fits into the frame, or:
  2. Everyone is extremely far from the camera, so 95% of the screen area is taken up by a conference table.

Figure 1 illustrates this common scenario.

conference-call-1-without-prism.png

Fig. 1: When multiple people are sharing a laptop during a conference call, usually the video looks like the example on right, where only one person is actually fully visible.

Proposal:

An inexpensive prism can fix this problem once and for all (Figure 2). A prism can be placed directly in front of the camera to split the image into multiple horizontally-spaced parts.

Now everyone can participate in the conference call without needing to move the camera around!

conference-call-2-with-prism

Fig. 2: The prism attachment makes it easy to fit everyone into frame. The prism could attach to the camera by means of either a magnetic clip or some sort of suction cup (probably the best solution for laptop screens).

PROS: Encourages conference call participation by people other than whoever happens to be directly in front of the camera.

CONS: Might result in an unflattering “fun house mirror” effect in the final image. (Although this could be fixed in software, or by a more complicated prism setup.)

Revolutionary new “lens-free camera” created in a garage by a crazy inventor—now all your vacation photos will be PERFECT!

Background:

When people go on vacation, 99% of their pictures are of sunsets and monuments that have been photographed thousands of times before.

Sure, that image of a majestic tropical bird perched in front of a waterfall may seem like one-in-a-million shot, but that still means that one hundred variants of it have already been uploaded to Google Photos.

The issue:

It can be a lot of work to frame a shot in an aesthetically-pleasing fashion. But what if we could make use of THE INTERNET to save us the trouble?

Proposal:

Instead of carrying a regular camera, a user can carry a “camera” without a lens or ability to take pictures.

Instead, when the user presses the shutter button:

  1. The “camera” records the user’s GPS coordinates, the time of day, and the current orientation of the camera.
  2. Later, the “camera” syncs this information to the Internet and downloads the most aesthetically-pleasing photo for the specified site and time of day.

So if you were disappointed that you were taking photos of Niagara Falls while it was overcast, no worries: the camera will pick out some majestic waterfall-and-rainbows-on-a-sunny-day photos from online.

take-a-photo

Fig. 1: This “lens free” camera looks like a traditional camera, except it doesn’t actually take pictures—it only records your GPS location and orientation when taking the picture. (The viewfinder shown here could just be a transparent plastic window, not a real LCD screen.)

 

it-actually-searches-online

Fig. 2: Internally, the camera is really just searching for the top-rated image at a given set of coordinate / orientation / time-of-day.

Conclusion:

Stop futzing with shutter speed, exposure, framing, and who knows what else—just let the Internet take your photos for you!

PROS: Saves tons of time! Makes everyone into a master photographer. You won’t have to worry about looking bad in a photo, because you’ll never be in a photo! (Unless you happen to be the subject of the top-rated photo somehow.)

CONS: At least one person will still have to use a real camera to take photos, or else there won’t be anything for the GPS-only camera to find.

 

 

ad-your-address-will-be-sold.png

Five underrated facts about dystopian totalitarian surveillance regimes! You’ll never believe fact #2!

Background:

The optimal tradeoff between privacy and security is a topic that is endlessly debated.

In the past, omnipresent surveillance was not feasible—but technology is now at the point where implementation of a 1984-esque surveillance state is actually possible.

On the one hand, it would be theoretically convenient to have immediate response to crimes and/or injuries, and perhaps take action to prevent some crimes before they even occur.

On the other hand, you might be sent to a faraway gulag because you opposed the interests of a politically-connected individual.

Proposal:

The problem here, of course, is the human element (see Figure 1).

monitor-computer-guy

Fig. 1: This guy (right) can monitor every aspect of your life on the video screens (left). This works fine until you become successful and he blackmails you!

But if an all-seeing computer system (like Skynet in the Terminator series) were in charge of things, we could could theoretically know that the surveillance system could not be misused, and would only be used for the programmed-in purposes (e.g., catching kidnappers and insane murderers).

Humans would write the rules for the system, but the raw data would (somehow) be inaccessible except to the analysis computer (Fig. 2).

Some example rules that might be applied:

  • If a car was used in a felony, check traffic cameras for its license plate number.
  • If a person has purchased explosive-manufacture-related chemicals, check their records for unusual activity and potentially flag them for further investigation by actual humans.
  • If a person declared no taxable income, but drives around in an 80,000 dollar car, check them for tax fraud.

Since these rules could be set by the legislature, they could be transparent and subject to review by the voters.

One downside: many countries operate on implicit rules like:

  • If a person supports an opposing political party, make sure to harass and imprison them.
  • If a person is a member of a disfavored ethnic or religious group, make sure to hold them to the strictest letter of the law.
  • Otherwise, don’t enforce any rules at all.

These informal enforcement rules might be less likely to survive if they had to be explicitly coded up and put on the official registry of surveillance rules. Or perhaps they would remain, and just be enforced with horrific robotic precision!

robot-wheel

Fig 2: This robot is totally trustworthy with your personal data, and has no ulterior motives or desires of its own (unlike a human).

seeing-eye

Fig 3: This unblinking “panopticon” eye will be a useful symbol to let you know you are in a safe and trustworthy robot-surveilled region! Stick one of these in your bedroom and bathroom to remind you that a robot is watching you at all times.

Conclusion:

When you lobby for omnipresent surveillance, make sure to imagine the predicable scenario where some irrationally angry neighbor or ambitious business rival now has a recording of every stupid thing you (and your friends/family) have ever done!

PROS: Would probably reduce many types of crime.

CONS: Terminator and/or 1984.