5 ms·

20 years ago I did some software to analyze satellite images of the Amazon to monitor deforestation. We got a result that matched the quality of human experts.

by neves 10y ago

20 years ago I did some software to analyze satellite images of the Amazon to monitor deforestation. We got a result that matched the quality of human experts. The problem has always been political and economical, not technological.

aabajian 10y ago
I agree, nevertheless it's exciting to see some progress being made (see the heart flow and iSchemaView companies cited below). Incidentally, there is a new angle to the radiologists push back - interventional radiologist are generally quite in favor of automating reads. This year IR is its own specialty, but coupled to three years of DR training. This is the speciality that I'm entering. I want to be an interventionalist while advocating for the adoption of machine learning systems on the diagnostic side.
spuz 10y ago
I'm not sure if there was more to your story that you missed. Was your software successful? Is it in use today? Is automation widespread in the study of Amazonian deforestation?
- tomsmeding 10y ago
  Software not being used all over the place does not mean the software is bad, generally.
- neves 10y ago
  No. No. No. :-)
i336_ 10y ago
How did your software work, what techniques was it using, and what was the surrounding context relating to the data that was fed in and the output the model gave? There might be some interesting things that can be learned from this kind of info and applied to the current status quo (I'm definitely not arguing that there is a sociopolitical element).
- neves 10y ago
  There were a bunch of papers: http://www.sciencedirect.com/science/article/pii/0888613X94900221 http://www.sciencedirect.com/science/article/pii/0888613X949...
profalseidol 10y ago
The new means of production has now been realized and the need to replace capitalism is now at hand. Hackers of the World Unite!
seanlinmt 10y ago
You are right that work in these areas tend not to be hampered by technology. It sometimes may even be hazardous to ones health considering the value of timber. 20 years ago is a long time ago, GIS must be still at its infancy then. These days one could pull free satellite images from NASA and probably could just diff images if not for those pesky clouds. Oh wait actually I think NASA does do earth imagery at various spectral ranges, https://www.odysseyofthemind.com/aster.htm https://www.odysseyofthemind.com/aster.htm. Curious to know what sort of methods you used then if you don't mind sharing.
- neves 10y ago
  Excuse me for the delay to answer. The best results were with backpropagation neural networks: http://www.sciencedirect.com/science/article/pii/0888613X94900221 http://www.sciencedirect.com/science/article/pii/0888613X949... But we also used fuzzy logic neural networks with genetic algorithms: http://ieeexplore.ieee.org/document/712156/?reload=true http://ieeexplore.ieee.org/document/712156/?reload=true It were 24 years ago.

by neves 10y ago

aabajian 10y ago

I agree, nevertheless it's exciting to see some progress being made (see the heart flow and iSchemaView companies cited below). Incidentally, there is a new angle to the radiologists push back - interventional radiologist are generally quite in favor of automating reads. This year IR is its own specialty, but coupled to three years of DR training. This is the speciality that I'm entering. I want to be an interventionalist while advocating for the adoption of machine learning systems on the diagnostic side.

spuz 10y ago

I'm not sure if there was more to your story that you missed. Was your software successful? Is it in use today? Is automation widespread in the study of Amazonian deforestation?

tomsmeding 10y ago
Software not being used all over the place does not mean the software is bad, generally.
neves 10y ago
No. No. No. :-)

i336_ 10y ago

How did your software work, what techniques was it using, and what was the surrounding context relating to the data that was fed in and the output the model gave? There might be some interesting things that can be learned from this kind of info and applied to the current status quo (I'm definitely not arguing that there is a sociopolitical element).

neves 10y ago
There were a bunch of papers: http://www.sciencedirect.com/science/article/pii/0888613X94900221 http://www.sciencedirect.com/science/article/pii/0888613X949...

profalseidol 10y ago

The new means of production has now been realized and the need to replace capitalism is now at hand. Hackers of the World Unite!

seanlinmt 10y ago

You are right that work in these areas tend not to be hampered by technology. It sometimes may even be hazardous to ones health considering the value of timber. 20 years ago is a long time ago, GIS must be still at its infancy then. These days one could pull free satellite images from NASA and probably could just diff images if not for those pesky clouds. Oh wait actually I think NASA does do earth imagery at various spectral ranges, https://www.odysseyofthemind.com/aster.htm https://www.odysseyofthemind.com/aster.htm. Curious to know what sort of methods you used then if you don't mind sharing.

neves 10y ago
Excuse me for the delay to answer. The best results were with backpropagation neural networks: http://www.sciencedirect.com/science/article/pii/0888613X94900221 http://www.sciencedirect.com/science/article/pii/0888613X949... But we also used fuzzy logic neural networks with genetic algorithms: http://ieeexplore.ieee.org/document/712156/?reload=true http://ieeexplore.ieee.org/document/712156/?reload=true It were 24 years ago.