search

KaihuaTang / Scene-Graph-Benchmark.pytorch

A new codebase for popular Scene Graph Generation methods (2020). Visualization & Scene Graph Extraction on custom images/datasets are provided. It's also a PyTorch implementation of paper “Unbiased Scene Graph Generation from Biased Training CVPR 2020”
MIT License
1.06k stars 228 forks source link

Detect Scene Graphs on Your Custom Images #91

Closed PipiZong closed 3 years ago

PipiZong commented 3 years ago

❓ Questions and Help

Hi,

I tried your command provided in the section, Detect Scene Graphs on Your Custom Images, to see what the scene graph looks like generated from 10 images sampled randomly from VG_100K ( i.e., 1.jpg-10.jpg from the dataset are taken as the custom images). After I visualized one of the scene graphs, I found the bounding box and its label, or the relationships were all predicted wrong. I am wondering if the way I did is wrong or scene graph detection is indeed inaccurate. Could you please give me some suggestions about generating a scene graph given an image without any annotations precisely?

Thanks!

kyteinsky commented 3 years ago

Hey @PipiZong, Can you share the modified code, something is not right when I changed the model weight path using MODEL.WEIGHT. It takes the /home/kaihua... path and nowhere can I see that nor it is overidden by command line argument.

I would be very grateful if you could share it by any medium suitable :smile:

Thank you.

rayush7 commented 3 years ago

@PipiZong I am facing the same problem. Did you manage to resolve the issue?

iz2late commented 3 years ago

Hey @PipiZong, Can you share the modified code, something is not right when I changed the model weight path using MODEL.WEIGHT. It takes the /home/kaihua... path and nowhere can I see that nor it is overidden by command line argument.

I would be very grateful if you could share it by any medium suitable 😄

Thank you.

I have the same problem :(