-
Thanks for your great work!
As mentioned in Appendix D.1,
> Considering that the number of objects in one image in COCO dataset ranges from 1 to 80, this design is inefficient and results in excessi…
fushh updated
2 years ago
-
I was recently trying to update a function in [TurbulenceConvection.jl](https://github.com/CliMA/TurbulenceConvection.jl), [here](https://github.com/CliMA/TurbulenceConvection.jl/pull/1285). In doing …
-
hello , I want to change the model into an class-aware detector, whichi part should I modify? Thank you !
-
I modified `projects/configs/denoise/petrv2_vovnet_gridmask_p4_800x320_dn.py` to generate a config file using ResNet-50DCN as backbone. After training, I got a lower performance compared to the paper:…
-
The number of Ground Truth is different in different images, so the number of noising queries can't be same for the images in a same batch, how do you solve this problem?
-
Nice work! However, the code seems not complete. For example, in the paper, you use retinaHead to replace RPNHead, but, you do not offer the fixed retinaHead. Hope you can update those incomplete cod…
-
Hi! I am using `mmdet` for a custom dataset. The object detection task I am working with is a single-class problem, and I am only concerned with predicting the bounding box for this class of objects. …
-
But I think this has always been a problem what exactly does specimen/event remark mean? It has never been clear.
Anyway, change this to "Collection Object Remark" to match data entry
![image](h…
-
### Issue Type
Others
### onnx2tf version number
1.9.2
### onnx version number
1.13.1
### tensorflow version number
2.12.0
### Download URL for ONNX
https://github.com/plemeri/InSPyReNet
###…
-
Hi zikun:
When will you plan to release the code? Looking forward to the GTELT.