SLAM

• run_CNN.py is for the initially proposed model using a step-wise learning approach.
• run_CNN.py might need several attempts to converge.

• run_CNN_RNN.py is for a new proposed model that uses LSTM cascaded to CNN.

Dataset Preparation:

For run_CNN.py

• /dataset contain directories for /training_set and /validation_set.
• Each contains sub-directories having multiple images of the same scene or class.
• The images belonging to same class or same scene have modified environmental conditions. (occultation, illumination, diff ViewPoints) conditions.
• The images may be scaled, rotated or partially occluded.

For run_CNN_RNN.py

• Create the following directories (for example):
• datasets -> GardensPointWalking-> day_left ; day_right ; night_right; (csv - optional)
• results
• checkpoints

Runtime Environment

For run_CNN.py

• The requirements.txt file list Python version and all libraries depend on: (if run on a local machine)
• Installed using:

pip install -r requirements.txt

• For Colab, use the below code to manage the versions.

import sys
print("User Current Version:-", sys.version)
!sudo apt-get update -y
!sudo apt-get install python3.8
!sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.8 2
!sudo apt-get install python3-pip
!sudo apt install python3.8-distutils
!python --version

• Further use the !pip instal command to install other dependencies given in requirements.txt

For run_CNN_RNN.py

• Directly run the bash command sh demo.sh
• Do the corresponding changes in the CLI bash file ./demo.sh: (local machine)
  • To change the model (Pre-train)
  • To change the number of input image sequences.
• For Colab (after changing the current (drive) directory)
• resnet18 as an example is used in demo.sh

!pip install tqdm
!pip install fire
!sh demo.sh