yqyao / fcos_plus Goto Github PK
View Code? Open in Web Editor NEWSome improvements (center sample) about FCOS (FCOS: Fully Convolutional One-Stage Object Detection).
License: Other
Some improvements (center sample) about FCOS (FCOS: Fully Convolutional One-Stage Object Detection).
License: Other
/home/Desktop/FCOS_PLUS-master/maskrcnn_benchmark/csrc/cpu/vision.h:3:29: fatal error: torch/extension.h: No such file or directory
#include <torch/extension.h>
我看到FCOS的两个版本的论文 都引用了你的 code 地址
https://github.com/aim-uofa/AdelaiDet/blob/master/configs/FCOS-Detection/README.md
Hello,
Thank you for your nice work.
Have you ever tried a model using ctr. sampling that does not have "centerness" branch?
As someone else mentioned in the issue, the centerness score of the sampled center points will be similar.
I wonder whether the "centerness" itself still gives an improvement.
Best,
Ahyun Seo
Traceback (most recent call last):
File "tools/train_net.py", line 182, in
main()
File "tools/train_net.py", line 175, in main
model = train(cfg, args.local_rank, args.distributed)
File "tools/train_net.py", line 81, in train
arguments,
File "/home/data/TeddyZhang/nothing/FCOS_PLUS/maskrcnn_benchmark/engine/trainer.py", line 57, in do_train
for iteration, (images, targets, _) in enumerate(data_loader, start_iter):
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 819, in next
return self._process_data(data)
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 846, in _process_data
data.reraise()
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torch/_utils.py", line 385, in reraise
raise self.exc_type(msg)
AttributeError: Caught AttributeError in DataLoader worker process 0.
Original Traceback (most recent call last):
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torch/utils/data/_utils/worker.py", line 178, in _worker_loop
data = fetcher.fetch(index)
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py", line 44, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/data/TeddyZhang/nothing/FCOS_PLUS/maskrcnn_benchmark/data/datasets/coco.py", line 67, in getitem
img, anno = super(COCODataset, self).getitem(idx)
File "/home/t704/Software/anaconda3/envs/Detectron/lib/python3.7/site-packages/torchvision/datasets/coco.py", line 118, in getitem
img, target = self.transforms(img, target)
File "/home/data/TeddyZhang/nothing/FCOS_PLUS/maskrcnn_benchmark/data/transforms/transforms.py", line 15, in call
image, target = t(image, target)
File "/home/data/TeddyZhang/nothing/FCOS_PLUS/maskrcnn_benchmark/data/transforms/transforms.py", line 60, in call
target = target.resize(image.size)
AttributeError: 'list' object has no attribute 'resize'
how to set the dataset ?? i set it by using maskrcnn-benchmark version... please help me !!
I change the centerness head to regression head, but found it isn't work in my experiment, is someone else did this experiment?
my datasets have 7 classes.
Thanks for your great work! The center sampling method is quite like the sampling method in FoveaBox, is that wright?
In many text detection or other anchor-free methods,when we shrink gt to get new "center location", the original zone of gt which not be chosen is set to "ignore". But when I read this code, I find it's set to 0 (neg) directly. do I misunderstand this code or IT IS?
many thanks for you!
I am little confused about the mutil-scale training, do you mean that use the crop or else to enhance the train set and use 2X at the same time? Did the original FCOS use this trick too?or just the 2X.
Can anyone , please share the trained models(.pth files) in google drive.
Why does center sampling bring so much improvement? Do you think training with center samples can lead to a better classification branch and regression branch in FCOS?
Moreover, center-ness scores of the center samples could be very close, so using center-ness during inference or not in FCOS_PLUS may lead to similar results?
I used thop,torchstat and ptflops to caculate the Flops of the model, but all of the results shows there is 0 params and 0 flops in ResNet and FPN. Why did this happen and how can I calucalte the Flops of model correctly. Thank you!
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.