Question about human performance

hellozhuo / pidinet

Code for the ICCV 2021 paper "Pixel Difference Networks for Efficient Edge Detection" (Oral).

Other

433 stars 65 forks source link

Question about human performance #25

Open ZhouCX117 opened 2 years ago

ZhouCX117 commented 2 years ago

Hi, could you please tell me how to compute the 0.803 ODS human performance? I can't find the corresponding reference. Thanks for your favor!

ZhouCX117 commented 2 years ago

Hi, could you give me a favor? I can't solve this problem.

zhuoinoulu commented 2 years ago

Hello, please see: Martin, D. R., Fowlkes, C. C., & Malik, J. (2004). Learning to detect natural image boundaries using local brightness, color, and texture cues. PAMI, 26(5), 530–549. and Y. Liu et al., "Richer Convolutional Features for Edge Detection," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, no. 8, pp. 1939-1946, 1 Aug. 2019, doi: 10.1109/TPAMI.2018.2878849.

ZhouCX117 commented 2 years ago

@zhuoinoulu Thanks for your answer! From the paper 1 you provided, the human performance is calculated by one human label to others. In our work, the model performance is calculated by one to many, and the human performance is many to many. Is it right? By the way, I could not design the code for calculating human performance.

zhuoinoulu commented 2 years ago

@ToBeNormal Sorry, I am not sure about that. since I just cited the number from the references. Please let me know if you figure it out.

ZhouCX117 commented 2 years ago

@zhuoinoulu Hi, I want to get your saved jpg during the training process. But I only see the logs, could you please provide the saved pictures? Thanks a lot!