-
Hi, at your [loss function](https://github.com/oxcsaml2019/multitask-learning/blob/master/multitask-learning/losses.py#L87) you calculate:
```python
loss += torch.exp(-sem_uncertainty) * sem_loss …
-
On the SUIM dataset (Semantic Segmentation of Underwater Imagery: Dataset and Benchmark), the performance of UNet (IoU:56.9700) is significantly lower than that of FCN(66.75) !!! In fact, the performa…
-
Hello everyone,
I am currently trying to add a CRF to the output of the mask graph to improve segmentation task after I saw this implementation of the referring paper: [https://github.com/Guruprade…
-
Hi!
I am trying to use the pretrained models to process images from KITTI Odometry and changed nothing of the code. But I got some invalid segmentations. Then I tested in the sample image4 in [here]…
-
First of all, thank you very much for your research.
I have some questions about the inference process. I used the checkpoint you trained for inference. Why did I report an error and the model parame…
-
Official implementation of the [Fully Convolutional Instance-aware Semantic Segmentation](https://arxiv.org/abs/1611.07709) (FCIS) based on MXNet -> they use hickle to dump masks fast. It is wotrh che…
-
On Unicorn, make it so that sem-seg and obj-det renderings are returned as separate entities. This is to be used during CSUN to get clearer feedback on how users are interpreting semantic segmentation…
-
I can't find the val result of 84.9%mIoU in the OCR paper. Is this the result reproduced by yours? If so ,can you describe your implementation details?
-
This happens for the language SME in the Giella infra. The regex file in question is reorder-semantic-tags.regex (attached).
```
$ time hfst-regexp2fst --verbose --format=sfst -S reorder-semantic-tag…
-
Thank you for your awesome work and open-source code!
Can you provide better documentation for the MEM stuff?
@mktk1117