• Developed the single object detection code to multilabel classification code and run the multilabel classification code

  • • • • Changed the "softmax" output layer into "sigmoid" output layer

        • Changed the loss from "categorical" into "binary_crossentropy"

Use below code to observe the Prediction precentage for each classes

cat = category_dict[cat_index]

print(f'{path}\t\tPrediction: {cat}\t{int(pred.max()*100)}% Confident')

classes={0:'Tire', 1:'Tin', 2:' Non'}

proba=loaded_model.predict(img.reshape(1,64,64,3))

top_3=np.argsort(proba[0]) [:-4:-1]

for i in range(3):

print("{}".format(classes[top_3[i]])+" ({:.3})".format(proba[0][top_3[i]]))

Next week target-Improve the code for multi object detection