IndexError with AnchorExplainer and Yolov8

[knowitall12]

Hi,

I am trying to get anchor explainer working for fine-tuned yolov8 model.

Error:

image 1/1 /content/content/dataset/images/valid/HCL_0_idx3_xs130_xe250_res416.png: 416x416 1 morning_star, 3632.7ms
Speed: 1.5ms preprocess, 3632.7ms inference, 1.0ms postprocess per image at shape (1, 3, 416, 416)

image 1/1 /content/content/dataset/images/valid/HCL_0_idx3_xs130_xe250_res416.png: 416x416 1 morning_star, 1993.1ms
Speed: 1.4ms preprocess, 1993.1ms inference, 1.0ms postprocess per image at shape (1, 3, 416, 416)

image 1/1 /content/content/dataset/images/valid/HCL_0_idx3_xs130_xe250_res416.png: 416x416 1 morning_star, 3030.0ms
Speed: 1.3ms preprocess, 3030.0ms inference, 0.9ms postprocess per image at shape (1, 3, 416, 416)

---

IndexError Traceback (most recent call last)
in <cell line: 17>()
15 explainer = AnchorImage(predict_fn, image_shape, segmentation_fn=superpixel)
16
---> 17 explanation = explainer.explain(image, threshold=.95, p_sample=.8, seed=0)
18 plt.imshow(explanation.anchor[:,:,0]);

4 frames
/usr/local/lib/python3.10/dist-packages/alibi/explainers/anchors/anchor_image.py in call(self, anchor, num_samples, compute_labels)
137 raw_data, data = self.perturbation(anchor[1], num_samples)
138 labels = self.compare_labels(raw_data)
--> 139 print(raw_data,labels,self.n_covered_ex)
140 covered_true = raw_data[labels][: self.n_covered_ex]
141 covered_true = [scale_image(img) for img in covered_true]

IndexError: boolean index did not match indexed array along dimension 0; dimension is 100 but corresponding boolean dimension is 1

Code:

import matplotlib
%matplotlib inline
import matplotlib.pyplot as plt
import numpy as np

model = YOLO("/content/final_model.pt")

image_path = "/content/content/dataset/images/valid/HCL_0_idx3_xs130_xe250_res416.png"
image = img = cv2.imread(image_path)
plt.imshow(image[:,:,0])
test_image = image.reshape(1, 416, 416, 3)
results = model([image_path])
print("Prediction ", results[0].boxes)

def superpixel(image, size=(16, 26)):
    segments = np.zeros([image.shape[0], image.shape[1]])
    row_idx, col_idx = np.where(segments == 0)
    for i, j in zip(row_idx, col_idx):
        segments[i, j] = int((image.shape[1]/size[1]) * (i//size[0]) + j//size[1])
    return segments

# trying with static probabilities
predict_fn = lambda x: np.array([0.01]*9+[0.82]+[0.01]*10).reshape(1,20)
image_shape = image.shape

explainer = AnchorImage(predict_fn, image_shape, segmentation_fn=superpixel)

explanation = explainer.explain(image, threshold=.95, p_sample=.8, seed=0)
plt.imshow(explanation.anchor[:,:,0]);