explanation = explainer(image, target_explain_group=TargetExplainGroup.ALL) fails with error: OpenCV(4.9.0) /io/opencv/modules/imgproc/src/resize.cpp:3789: error: (-215:Assertion failed) !dsize.empty() in function 'resize'
TODO:
Update README: update structure, add reciprocam arxiv link
Substitute loading tests/assets/cheetah_person.jpg image with wget link (can be done during PR to OV_notebooks)
Fix minor text issues
Update pypl modules for numpy 2.0 to compile
Check that target_layer for MobiilenetV3 is correct. It returns correct values, but it's not the last conv layer
Question:
Do we need Advanced section and Insert XAI branch section
There's duplicated code
# Check saved saliency maps
print(f"Saliency maps were generated for the following classes: {list(explanation.saliency_map.keys())}")
print(f"Saliency map size: {explanation.shape}")
What was done:
Issues:
explanation = explainer(image, target_explain_group=TargetExplainGroup.ALL)
fails witherror: OpenCV(4.9.0) /io/opencv/modules/imgproc/src/resize.cpp:3789: error: (-215:Assertion failed) !dsize.empty() in function 'resize'
TODO:
tests/assets/cheetah_person.jpg
image with wget link (can be done during PR to OV_notebooks)target_layer
for MobiilenetV3 is correct. It returns correct values, but it's not the last conv layerQuestion:
Advanced section
andInsert XAI branch
sectionShow saliency maps for cheetah class
cheetah_raw_sal_map = explanation.saliency_map[cheetah_class_index] cheetah_raw_sal_map = cv2.cvtColor(cheetah_raw_sal_map, cv2.COLOR_BGR2RGB)
plt.imshow(cheetah_raw_sal_map) plt.show()