Implementing self driving car in NFS:MW 2012 sandbox environment using tflearn-Alexnet CNN ,lane detection
TRAINED AND TESTED ON PYTHON 3.7 & TENSORFLOW 2.5
latest_ver_2...
Currently in ver_2 , this model behaves better on highways and wide roads with proper lane markings.
Accuracy = 93.6 %
Validation accuracy = 94.73 % AFTER 15 EPOCHS (Due to resource limitations)
RAW FOOTAGE to check how this model behaves: https://youtu.be/_uycuLdZTN4 https://youtu.be/NkuiH1UMiHs
Training data & Model : https://1drv.ms/u/s!AizZTWOt5j3cgXguYPDRqzz6hMU_?e=tTSs4l
Untitled.mp4
(ver_1) Implementing self driving car in NFS:MW 2012 sandbox environment using tflearn-Alexnet CNN ,lane detection
This is version_1.00 of this project
Accuracy = 91.64 % Validation accuracy = 91.80 % ... this is calculated after 10 epochs.
-> for generating training data , use "training_data_collector.py" from "latest_july21" folder.
-> input frames are captured from LEFT-TOP corner of the screen in resolution '800 x 600'.
Untitled.mp4
