This repo contains projects on implementaion of different object detection algorithm.

• Install python 3.6.2 64-bit version or Newest Version
• Run these commands :

  pip3 install tensorflow==1.14
  pip3 install tensorflow-gpu==1.14 (Stable but visual C++ 2015 v3 update required)
  

• Install cuda 10
• Download cudNN from Nvidia after Login.
• Copy contents of cuDNN 10 to C:\Program Files\NVIDIA GPU Computing Toolkit. I have used cudNN v11.
• You might need other cudNN downloads to copy-paste dll files.

You may find several missing dll. Just find them on internet or go to C:\Program Files\NVIDIA GPU Computing Toolkit. Find the similar dll and rename it.

This is the chart to make accurate versioning between tensorflow GPU, cuda and cudNN

StackOverFlow | TensorFlow

To check if your GPU is enlisted:

import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0' # 0 = GPU use; -1 = CPU use
from tensorflow.python.client import device_lib
print(device_lib.list_local_devices())

import tensorflow as tf
print("Num GPUs Available: ", len(tf.config.experimental.list_physical_devices('GPU')))

if tf.test.gpu_device_name():
    print('GPU found')
else:
    print("No GPU found")

Tutorial from Medium

import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0' # 0 = GPU use; -1 = CPU use

import keras
import tensorflow as tf

config = tf.compat.v1.ConfigProto( device_count = {'GPU': 1 , 'CPU': 3} )
sess = tf.compat.v1.Session(config=config)
keras.backend.set_session(sess)

If you face problem with 'import keras.something' convert it to'tensorflow.python.keras.something'

import tensorflow as tf
print(tf.version.VERSION)

Download MatterPort Github repo on MaskRCNN Use it according to your need. Read README.MD

Tutorial Link from YouTube

Most cruicial. No good update for windows 10. Visit here

Yolov3 | Yolov3 Tutorial with direct weight | Convert to Coco Format

Project Explanation

• A very details guideline on Mask-RCNN is published in Medium by the Author Ahmed Yunus.
• However to understand YOLO, this link is referred.
• The codes are self explanatory. The function names are completely aligned with what those are meant to do.