kaixin96 / PANet

Code for our ICCV 2019 paper PANet: Few-Shot Image Semantic Segmentation with Prototype Alignment
322 stars 64 forks source link

Model Computation Time #14

Closed AlgorithmicIntelligence closed 4 years ago

AlgorithmicIntelligence commented 4 years ago

I evaluate the processing time on inference with the line of code as below. ''' time_s = time.time() querypred, = model(support_images, support_fg_mask, support_bg_mask, query_images) time_elapsed = time.time() - time_s ''' I just found that the FPS of the model can above 30, is that really? Because I though instance segmentation usually time-consuming.

kaixin96 commented 4 years ago

Hi @AlgorithmicIntelligence , I just checked previous testing log. For 1-way 1-shot setting, the wall clock time for evaluating on 1000 episodes (on a single Titan X) is 5 mins and for 1-way 5-shot it is 8 mins. Hope this information helps.

I did not measure the FPS of our model but 30 is too fast to be true. By the way, our work is in semantic segmentation category instead of instance segmentation.

Thank you :)

kaixin96 commented 4 years ago

I’m closing this issue because it has been inactive for a few months. Feel free to reopen if you have questions. Thank you.