-
I created a simple model that adds a global pooling layer and then 2 dense layers to VGG16 and learns to discriminate between 3 classes. It goes like this:
```
base_model = VGG16(weights='imagenet…
-
Hi,
I would like to retrain the model and am interested in understanding how to measure and display the metric values such as Correlation Coefficient (CC), Structural Similarity Index (SIM), Normal…
-
hi, may i ask how to get the mask in the DefectMaker part, When i tested items[mask], there is nothing, and mask_size is None,
-
Something wrong at the step of "Get perturbation saliency map"
I found "visualize_atari/jacobian-vs-perturbation.ipynb" still cannot work through, at the step of "Get perturbation saliency map":
`…
daiyl updated
4 years ago
-
Hi,
Why do we need saliency maps to train the model?
Where can we find the saliency maps for ImageNet dataset?
Thanks
-
Running the example for vanilla gradient caused error:
`
TypeError Traceback (most recent call last)
in ()
58 from saliency import GradientSaliency
---> 59 v…
-
Hi,
Firstly, thanks for your contribution. I have 2 GPUs system with RTX3080ti (12 GB RAM) when i run train.sh i got CUDA error. i reduced batch size but i got same error again, how can i resolve?
…
-
The input model (including weights) and image are exactly the same, but visualize_saliency showed different results.
(1). Use "pip install keras-vis" to install today
>>> mapp = visualize_salie…
-
Awesome repo; thank you for creating this! I was wondering if you looked into visualizing the Grad CAM saliency maps for the ViT based CLIP models (and if so, which layer we should select as the salie…
g-luo updated
2 years ago
-
[Saliency maps (or sensitivity masks)](https://pair-code.github.io/saliency/) visualize which inputs contribute towards a model decision. They are generated from gradient-based methods such as [integr…