Auto Count is already very strong, but one improvement could materially increase accuracy for electrical, fire alarm, security, HVAC, and other symbol-heavy trades: recognize the individual elements within a compound symbol and use them to prevent duplicate counting.
Example:
A drawing may contain:
Today, if I Auto Count the combined WF/TS symbol and later Auto Count standalone TS symbols, the TS portion inside the combined symbol will also be picked up—creating a duplicate.
Ideally, Auto Count would understand:
WF/TS = WF element + TS element
Then, when searching for standalone TS, it would recognize that a matching TS already exists as an element within a previously counted compound symbol and exclude it.
The key concept is leveraging the individual vector/image elements that make up each symbol, not just treating the symbol as one flat image.
For example, a combined WF/TS symbol may be made up of roughly 10 individual elements, while the standalone TS may consist of only 5 of those elements. A smarter recognition engine could use that relationship to understand that the TS is a subset of the larger WF/TS symbol.
A useful workflow would be:
This would effectively create a hierarchy of symbol matches where the more complete element match takes priority over a partial/subset match.
There may be additional logic required but the important concept is:
Use symbol recognition + element-level comparison + match priority/reassignment to prevent duplicate counts between standalone and compound symbols.
That would significantly improve Auto Count accuracy for trades where symbols frequently share common sub-elements.
Also - Seperate item (Useful Nice to have but not critical). In lieu of applying a symbol overlay (star, square, etc.) on top of the counted items, have the individual symbol's geometry highlighted. It is much easier to to quickly verify what has been counted, if duplicates exist, and what still remains.
Thank you for the extremely detailed feature request! We've been discussing improvements to Auto Count internally, and the de-duping would be a great addition. 🙂
👍1