adding explanations and similar samples

closest_sample.py

@@ -0,0 +1,42 @@
+from sklearn.decomposition import PCA
+import pickle as pk
+import numpy as np
+import pandas as pd
+
+
+
+pca_fossils = pk.load(open('pca_fossils_170_finer.pkl','rb'))
+pca_leaves = pk.load(open('pca_leaves_170_finer.pkl','rb'))
+embedding_fossils = np.load('embedding_fossils_170_finer.npy')
+#embedding_leaves = np.load('embedding_leaves.npy')
+
+fossils_pd= pd.read_csv('fossils_paths.csv')
+
+def pca_distance(pca,sample,embedding):
+    s = pca.transform(sample.reshape(1,-1))
+    all = pca.transform(embedding[:,-1])
+    distances = np.linalg.norm(all - s, axis=1)
+    print(distances)
+    return np.argsort(distances)[:5]
+
+def return_paths(argsorted,files):
+    paths= []
+    for i in argsorted:
+        paths.append(files[i])
+    return paths
+
+
+def get_images(embedding):
+    
+    #pca_embedding_fossils = pca_fossils.transform(embedding_fossils[:,-1])
+    
+    pca_d =pca_distance(pca_fossils,embedding,embedding_fossils)
+    
+    fossils_paths = fossils_pd['file_name'].values
+    
+    paths = return_paths(pca_d,fossils_paths)
+    print(paths)
+    paths= [path.replace('/gpfs/data/tserre/irodri15/Fossils/new_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/512/full/jpg/',
+                         '/media/data_cifs/projects/prj_fossils/data/processed_data/leavesdb-v1_1/images/Fossil/Florissant_Fossil/original/full/jpg/') for path in paths]
+    print(paths)
+    return paths 

explanations.py

@@ -0,0 +1,80 @@
+import xplique
+import tensorflow as tf
+from xplique.attributions import (Saliency, GradientInput, IntegratedGradients, SmoothGrad, VarGrad,
+                                  SquareGrad, GradCAM, Occlusion, Rise, GuidedBackprop,
+                                  GradCAMPP, Lime, KernelShap)
+
+import numpy as np
+import matplotlib.pyplot as plt
+from inference_resnet import inference_resnet_finer, preprocess, _clever_crop
+BATCH_SIZE = 1
+
+def show(img, p=False, **kwargs):
+    img = np.array(img, dtype=np.float32)
+
+    # check if channel first
+    if img.shape[0] == 1:
+        img = img[0]
+
+    # check if cmap
+    if img.shape[-1] == 1:
+        img = img[:,:,0]
+    elif img.shape[-1] == 3:
+        img = img[:,:,::-1]
+
+    # normalize
+    if img.max() > 1 or img.min() < 0:
+        img -= img.min(); img/=img.max()
+    # check if clip percentile
+    if p is not False:
+        img = np.clip(img, np.percentile(img, p), np.percentile(img, 100-p))
+    plt.imshow(img, **kwargs)
+    plt.axis('off')
+    
+    #return img
+    
+
+
+def explain(model, input_image,size=600, n_classes=171) :
+    """
+    Generate explanations for a given model and dataset.
+    :param model: The model to explain.
+    :param X: The dataset.
+    :param Y: The labels.
+    :param explainer: The explainer to use.
+    :param batch_size: The batch size to use.
+    :return: The explanations.
+    """
+    
+    # we only need the classification part of the model
+    class_model = tf.keras.Model(model.input, model.output[1]) 
+    
+    explainers = [
+             Saliency(class_model),
+             IntegratedGradients(class_model, steps=50, batch_size=BATCH_SIZE),
+             SmoothGrad(class_model, nb_samples=50, batch_size=BATCH_SIZE),
+             #GradCAM(class_model),
+  ]
+    cropped,repetitions = _clever_crop(input_image,(size,size))
+    size_repetitions = int(size//(repetitions.numpy()+1))
+    X = preprocess(cropped,size=size)
+    Y = np.argmax(class_model.predict(np.array([X])))
+    X = np.expand_dims(X, 0)
+    explanations = []
+    for e,explainer in enumerate(explainers):
+        print(f'{e}/{len(explainers)}')
+        Y = tf.one_hot([Y], n_classes)
+        phi = np.abs(explainer(X, Y))[0]
+
+        if len(phi.shape) == 3:
+            phi = np.mean(phi, -1)
+        
+        show(X[0][:,size_repetitions:2*size_repetitions,:])
+        show(phi[:,size_repetitions:2*size_repetitions], p=1, alpha=0.4)
+        
+        plt.savefig(f'phi_{e}.png')
+        explanations.append(f'phi_{e}.png')
+      
+    print('Done')  
+    
+    return explanations

pca_fossils_170_finer.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c622f5bb61b70fed1de24ad012524ee126fee662c7fe019bb966b6eea5b6922
+size 27685

pca_leaves_170_finer.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a43b8b1e1998f01fdf5f4a05458b9884fdd3ac9c3fa7e01875ca4e5e48078879
+size 27685