Taking the leap from 2D towards 3D with a 70+year old company?

[Abbyjay]

Hey guys, It's been a while, first happy new year 2018


I have a huge question and I don't really know where to start. My company still works actively with old Tif files for they are the only copies they have for their original structural equipments. Time is obviously gettingt o these old drawing for very little of them are scanned...GASP!!! yes my horror to work with. Recently my information management team heard about IBM Watson and decided we want this.


How do we passe from .tif and .dwg to 3D and aim for such a technology?

Sorry if this question doesn't suit this area.


Thanks

[ReMark]

What is the significance of mentioning IBM Watson?

If you really want to go from .tif to .dwg then take a look at the SCAN2CAD program.

As for 3D, this capability is already part of full AutoCAD so I'm not sure what you are getting at. Perhaps your company should be considering a MCAD program such as Inventor or Solidworks.

[Abbyjay]

Thanks Remark, I do work Inventor when ever I have the oppertunity to create new designs rather than simple updates or red lines. Too much of our valuable information is still paper format in cabinets. Softwares have been implanted previously claiming to be what we need and the company realised it is getting us nowhere. Now the leap is to figure out if we scan to cad, or laser scan everything to start from scratch. The mention of IBM Watson is the vision the company wants, to have all the information on any of our sites equipments accessible to anyone in the field instead of having to constantly contact the information management department then consuming time to explain the request,then search, then send the results. Upper management want to see results and my role is to assist in the first step of where to start...

[ReMark]

So are you personally taking the lead on this? If so, what is the first problem you want to solve?

[f700es]

Hire people to "jump in" and start recreating those older designs in 3D with whatever program you decide to go forward with. There is NO magic solution.

[ReMark]

I still don't see the rational behind the goal of employing IBM Watson to manage and share information between the company and the reps in the field. What is the benefit over any other well recognized and widely used method?

[Abbyjay]

IBM is what caught the upper managements heart now us under team have to make miracles...fast.! Thanks guys for your input on this.

[ReMark]

IBM Watson is by no means a miracle worker. Your company is going to have to put some real effort into this rather than just go through the motions. I sense that no one is really taking the lead on this effort and it too will be doomed to failure. Get back to us a year from now and let us know what progress you've made.

[CyberAngel]

As others have noted, there is no magical solution for converting from scanned paper to digital. The fact that your company still has analog resources tells me they've fallen far behind the curve.

The good news is that once you start to convert to digital, you can build some tools to accelerate the process. Look for recurring patterns and define blocks. Cut and paste wherever you can. Insert the image files and trace them when you must. Unfortunately, AI (such as Watson) still doesn't do well at visual recognition, something we take for granted.

For instance, title blocks are a good place to start. Trace or create a block. Give it some attributes for fields that vary from project to project. Put it in a template that you can use for all the subsequent conversions.

[lrm]

If you have .tif files that implies the drawings have already been scanned. As you know raster image files like tif, jpg, gif, etc., are “dumb” representations of geometry. They do not include a geometric model as found in CAD files. I get the impression that the reference to Watson is a way of asking if modern “intelligent” software can automatically convert a raster image file to a CAD database. There are commercial products today that can convert raster to vector (line) files with some, but not always complete, success. Even after the conversion the accuracy of the resulting CAD file is dubious.

I think the firs step for the company is to have scanned raster files of all drawings that may be referenced in the future. New projects should of course be doe with CAD. Old parts for which the documentation only exists as scanned images can be manually converted to CAD on an as-needed basis.

Note, the AI and image recognition aspects of the technology available with Watson and other similar products may be able to convert scanned images into well constructed CAD model in the future but when is anybody’s guess. The methods used to do face recognition and other types of image processing have made amazing strides with impressive results in the past few years. I was recently impressed by the ability of a friend’s new car to be able to “read” a speed limit sign as we drove down the highway. This is achieved by sophisticated signal processing that is useful in determining whether a collection of dots in a raster image is a line or not. The challenge is how the computer can determine that a collection of lines forms a 2D view and that a collection of 2D views define a 3D object with dimensions specified by dimensioning annotation in the scanned image.