It had become common for AI to work from pictures..
The best pictures you’ll give to AI, the more efficient the algorithms will be.
Here is one of the most useful case of scientific photography.
If you feed a computer with standardized pictures, it will process it faster with the best precision possible. Computers and humans “see” pictures on a different ways. Each human see a same subject differently. Our eyes are differents, our brain are differents. Also, we interpret a specific colors depending of the surroundings ones. Our mind can be tricked, this is why optical illusions exist.

In addition to that, the human eye is not very efficient to distinguish multiple colors. We are able to recognize approximately 4 millions colors.

A computer doesn’t “see” the whole picture.
It will know how many pixels there are and what is the color of each pixel with 3 number : one per channel RGB. For a computer, black will be 0 of Red, 0 of Green, 0 of Blue.

The One Health logo blue color is 77 Red 142 Green and 201 Blue.
RGB can be “8 bits” which means that each channel will have 2^8 colors possible = 256. 8 bits RGB will display 256x256x256= 16,7 millions colors. Its possible to shoot on 8, 10, 12 and 14 bits per channel. More than a human eye can see. But an AI can discriminate a infinity of colors because it works with numbers. Human eye could be able to see the difference between RGB 100,200,100 and 101,200,100 but AI will between 10000,20000,10000 and 10001,20000,10000..

Another weakness of the human vision is that we depend of the technology that display images.
When you want to display a color on a screen, computer sends an order to the screen “display on this pixel the color 100,200,100”.

Depending of the technology used to build the screen (and his factory calibration), the screen can be wrong and display 101,200,100.

The most expensive screens can display reliable 12 bits information but most of the common screens struggle with 8 bits.

You can test your color vision here :

Try this test with multiple screens (your smartphone, computer, tv screen…)

AI’s algorithm can benefit for 14 bits pictures to recognize patterns the human eye will never see. This is why a scientific picture must be think depending of the usage of the picture.

3D reconstruction and 3D printing

A computer is able to produce a 3D reconstruction from 2D pictures. With enough photographs, programs like Agisoft Photoscan will create both geometric and textured model that can be use for virtual modeling and 3D printing.
These software will benefit for specific settings and shooting protocols, reducing the post processing stage that can be time and skill consuming.
In a general way, the more you think and prepare your photograph, the less work you’ll have after taking the shot.


More and more cameras can record the GPS position of the shot, in addition to the date... Within a single picture, you can show a specific disease and its location. It’s simplify the data collection and decrease the human error.

Atlas and collection numerisation

Standardized pictures can be used for anatomy atlas for example. Taking all the photographs the same way allow a trustfull comparison between species or healthy/sick specimens. Lien vers cytomine phoque
A lot of institutions own magnificient collections of animals, insects and plants. These collections are only accessible if you go see them. Behind glass walls, you can’t observe all the angles you’d want to. Taking high resolution standardized shot allow the sharing of this intellectual content worldwide.
Generate the pictures you need for teaching in your own institution. Pictures can be obtained by the students. Students learn to take pictures, and these pictures can be used to train the students ! You choose what you want to show, and these pictures can be shared if needed.

Example here :PHOTOPSY


Teaching generalist to take standardized pictures of skin disease so a distant dermatologist will be able to recognize and trust the photograph.
This will permit to extend telemedicine for multiple lesions and specialties.
is a very good choice for collaborative work around scientific pictures.
Cytomine is a very good choice for collaborative work around scientific pictures.