Problem_1 - Clustering and Segmentation -- https://github.com/rahul-lyukbot/Assesments/blob/main/Softsensor_Assesment_1.ipynb
output :- https://github.com/rahul-lyukbot/Assesments/blob/main/assesment1_output.csv
Problem_2 - Process and analyze the public data provided by SBA.gov for each US state toextract meaningful insights from features in the dataset. https://github.com/rahul-lyukbot/Assesments/blob/main/Softsensor_problem_2.ipynb
- 1. converting image into tensors https://github.com/rahul-lyukbot/Assesments/blob/main/Softsensor_problem_3.ipynb
- 2. Linear filtering https://github.com/rahul-lyukbot/Assesments/blob/main/Softsensor_problem_3.ipynb
- 3. 3.ICA(Independent Component Analysis) https://github.com/rahul-lyukbot/Assesments/blob/main/Softsensor_problem_3.ipynb
- 4. Pixelation https://github.com/rahul-lyukbot/Assesments/blob/main/4_Pixelation.py
- 5. Image restoration https://github.com/rahul-lyukbot/Assesments/blob/main/5_Image%20restoration.py
Which filter will you prefer for the detect and visualization of catheters and lines in X-ray images and why?
Most efficient preprocessiing for Deep_learning is converting image to tensor and then genrate the data.this is because finding pattern in preprocess tenosors is very effective and loading is very fast due fetch and prefetch
How can deep learning help to detect presence of catheters and lines in images and which model you prefer?
best performacne we got on these model:-
- efficientnet_v2 https://tfhub.dev/google/collections/efficientnet_v2/1
- imagenet/mobilenet_v2_075_128/feature_vector https://tfhub.dev/google/imagenet/mobilenet_v2_075_128/feature_vector/5