AutoCAD 2025 includes a technology preview that uses machine learning to scan your drawing for objects that can be converted into blocks.
What is a Tech Preview?
When you start Object Detection, the drawing's geometry is sent to the AutoCAD machine learning service for recognition. Once the service has analyzed the drawing, the palette will notify you if any objects have been detected that can be converted to blocks. To further inspect and assess the identified instances, click Review Objects.
In review mode, the Detection toolbar is displayed at the top of the drawing window. This toolbar provides various options for navigating and managing detection results. The detection results are grouped into sets of similar objects. A
set represents a group of similar detected objects that can be converted into instances of a single block.
Within each set, the
primary instance (outlined by a blue border) serves as the block definition when converting into a new block. Alternatively, if you choose to convert into an existing block, the primary instance is used to define the insertion scale and rotation.
You may encounter incorrect sets or instances during detection review. In the following example, a similar object is not detected and included in the set. You can report this error using the Detection toolbar. By reporting errors, you contribute to improving the overall performance and accuracy of the machine learning service.
Key Considerations
Object Detection is a technology preview, which means that the detection capability and functionality will be further enhanced and improved over time.
It's important to note that Object Detection is optimized for plan-view drawings. The types of objects that AutoCAD can detect are continuously expanding. Currently, Object Detection is best at recognizing the following objects:
- Single-swing doors
- Double-swing doors
- Toilets
- Bathtubs
- Urinals
- Generic lighting symbols
Note: Object Detection relies on machine learning algorithms, so the results may not always be completely accurate or comprehensive.
To optimize detection accuracy as we continue to improve the machine learning algorithm, it's recommended to set your drawing to real-world units before you start the detection.
While in review mode, the following commands are temporarily blocked:
- BEDIT
- COMPARE
- COUNT
- COUNTLIST
- PLOT
- PUBLISH
- REFEDIT
- TRACE
- XCOMPARE
New Commands
DETECT - Displays the Detection palette and starts object detection.
DETECTCLOSE - Ends the detection review.
DETECTCONVERT - Displays the Convert dialog box, which you can use to convert the highlighted instances into blocks.
DETECTPRIMARY - Specifies the detected instance from which a new block definition is created.
DETECTREMOVE - Removes selected instances from the set.
DETECTREVIEWNEXT - Displays the next set of detected objects.
DETECTREVIEWPREV - Displays the previous set of detected objects.
