henrylee2570 / noisyanchor Goto Github PK
View Code? Open in Web Editor NEWLearning from Noisy Anchors for One-stage Object Detection
License: Apache License 2.0
Learning from Noisy Anchors for One-stage Object Detection
License: Apache License 2.0
Hi there. Thanks for very much for making your code public!
I've been experimenting with this repository and have encountered an error during training on a custom dataset. Namely, after a few hundred iterations, I got the following error:
ERROR [04/12 14:58:31 d2.engine.train_loop]: Exception during training:
Traceback (most recent call last):
File "/home/root/NoisyAnchor/detectron2/engine/train_loop.py", line 132, in train
self.run_step()
File "/home/root/NoisyAnchor/detectron2/engine/train_loop.py", line 215, in run_step
loss_dict = self.model(data)
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/root/NoisyAnchor/detectron2/modeling/meta_arch/retinanet_noisy.py", line 220, in forward
gt_classes, gt_anchors_reg_deltas, soft_labels, reweight_coeffs = self.get_ground_truth(anchors, gt_instances, box_cls, box_delta)
File "/usr/local/lib/python3.8/dist-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "/home/root/NoisyAnchor/detectron2/modeling/meta_arch/retinanet_noisy.py", line 364, in get_ground_truth
gt_matched_idxs, anchor_labels, topN_inds = self.matcher(match_quality_matrix)
ValueError: not enough values to unpack (expected 3, got 2)
Digging a bit deeper, I noticed that when there are no ground truth boxes in an example, only two objects are returned when calling MatcherTopN
(see here). However, three objects are expected in the training loop (see here).
I'm guessing some kind of default_topN_indices
needs to be returned, though I'm not sure what would be most appropriate.
Thanks!
Thank you for your work. I have not used detectron, could you please tell me the file location of this Noisy Anchor implementation?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.