search

norlab-ulaval / PercepTreeV1

Implementation of Grondin et al. 2022 "Tree Detection and Diameter Estimation Based on Deep Learning". Also includes datasets and some of the pretrained models.
Apache License 2.0
90 stars 17 forks source link

Diameter estimation #10

Open Ellingie opened 6 months ago

Ellingie commented 6 months ago

Hi,

The program works very good with the test data and my own, spectacular! I can't figure out how to transform the diameter in pixels to the real life diameter however. Could you help me with this? I have the pixels of every tree from every frame in an array now but i can't fix the tranformation.

Thanks in advance !

Greets, Elias

anhelina007 commented 6 months ago

Hi Elias, could you please share your requirements.txt file? I might help you, but in my case I have pycoco errors related to json file I am using

sdmsr commented 6 months ago

hello anyone can help with the diameter estimation because i cant find a way to get the results. Thank you

VGrondin commented 5 months ago

Hi guys, happy to know the models are useful!

Sorry for the delay, I don't get issues notifications from github since norlab-ulaval is the owner. To convert the diameter from pixel units into metric units, you will need depth images (or point cloud measurements). Once you have the depth value for your diameter keypoints, you can "easily" convert it to metric. Hope this helps

sdmsr commented 5 months ago

Thank you. I just need an information about the specs of the ZED stereocam used for the Canatree100 dataset mainly THE FOCAL LENGTH AND SENSOR WIDTH because i am trying to perform the diameter estimation on this dataset because i dont have a stereocam of my own to take new measurements. Thank you