Feature Visualization
How neural networks build up their understanding of images
https://distill.pub/2017/feature-visualization/
#CNN #convolutional_neutral_network
#Visualization
How neural networks build up their understanding of images
https://distill.pub/2017/feature-visualization/
#CNN #convolutional_neutral_network
#Visualization
Distill
Feature Visualization
How neural networks build up their understanding of images
#آموزش
Keras and Convolutional Neural Networks (CNNs)
https://www.pyimagesearch.com/2018/04/16/keras-and-convolutional-neural-networks-cnns/
#Keras #CNN
Keras and Convolutional Neural Networks (CNNs)
https://www.pyimagesearch.com/2018/04/16/keras-and-convolutional-neural-networks-cnns/
#Keras #CNN
PyImageSearch
Keras and Convolutional Neural Networks (CNNs) - PyImageSearch
This gentle guide will show you how to implement, train, and evaluate your first Convolutional Neural Network (CNN) with Keras and deep learning.
#مقاله #سورس_کد
Finding Tiny Faces (CVPR 2017)
pic: https://github.com/cydonia999/Tiny_Faces_in_Tensorflow/raw/master/images/selfie.jpg
🕐Runtime efficiency:
Resnet101-based detector runs at 1.4FPS on
1080p resolution and 3.1FPS on 720p resolution.
🔗abstract:
https://arxiv.org/abs/1612.04402
🔗Paper:
https://arxiv.org/pdf/1612.04402
🔗Project Page:
https://www.cs.cmu.edu/~peiyunh/tiny/
🔗codes:
⚙️#Matlab / #matconvnet implementation :
https://github.com/peiyunh/tiny
⚙️#Python / #Tensorflow implementation :
https://github.com/cydonia999/Tiny_Faces_in_Tensorflow
مرتبط با مقاله mtcnn
https://t.me/cvision/249
#face #face_detection
#deep_learning #cnn #convolutional_neutral_network
Finding Tiny Faces (CVPR 2017)
pic: https://github.com/cydonia999/Tiny_Faces_in_Tensorflow/raw/master/images/selfie.jpg
🕐Runtime efficiency:
Resnet101-based detector runs at 1.4FPS on
1080p resolution and 3.1FPS on 720p resolution.
🔗abstract:
https://arxiv.org/abs/1612.04402
🔗Paper:
https://arxiv.org/pdf/1612.04402
🔗Project Page:
https://www.cs.cmu.edu/~peiyunh/tiny/
🔗codes:
⚙️#Matlab / #matconvnet implementation :
https://github.com/peiyunh/tiny
⚙️#Python / #Tensorflow implementation :
https://github.com/cydonia999/Tiny_Faces_in_Tensorflow
مرتبط با مقاله mtcnn
https://t.me/cvision/249
#face #face_detection
#deep_learning #cnn #convolutional_neutral_network
#مقاله
این مقاله در حال حاضر بهترین الگوریتم تشخیص چهره در چالش WIDER است.
PyramidBox: A Context-assisted Single Shot Face Detector
https://arxiv.org/abs/1803.07737
#PyramidBox achieves superior performance among the state-of-the-art on the two common face detection benchmarks, #FDDB and #WIDER FACE.
#face #face_detection #deep_learning #cnn #convolutional_neutral_network
این مقاله در حال حاضر بهترین الگوریتم تشخیص چهره در چالش WIDER است.
PyramidBox: A Context-assisted Single Shot Face Detector
https://arxiv.org/abs/1803.07737
#PyramidBox achieves superior performance among the state-of-the-art on the two common face detection benchmarks, #FDDB and #WIDER FACE.
#face #face_detection #deep_learning #cnn #convolutional_neutral_network
#خبر
فیس بوک با میلیاردها تصویر آپلود شده در اینستاگرام یک شبکه کانولوشنالی را آموزش داد!
فیس بوک اخیرا با استفاده از 3.5 میلیارد تصویر اینستاگرام، در 17000 هشتگ مختلف یک شبکه عصبی کانولوشنالی را آموزش داده است.
این شبکه بر روی داده های تست چالش ImageNet به دقت 85.4 دست یافته است، که بهترین مدل قبل از این به دقت 83.1 دست یافته بود!
نکته قابل تامل نویزی بودن هشتگهای اینستاگرام در مقابل مجموعه داده ی بادقت برچسب زده شده ی آموزشی imagenet است.
به عقیده فیس بوک:
"As training data sets get larger, the need for weakly supervised — and, in the longer term, unsupervised — learning will become increasingly vital," said Facebook. "Understanding how to offset the disadvantages of noisier, less curated labels is critical to building and using larger-scale training sets."
https://www.engadget.com/2018/05/02/facebook-trained-image-recognition-ai-instagram-pics/
https://www.wired.com/story/your-instagram-dogs-and-cats-are-training-facebooks-ai/
#classification #image_recognition
#convolutional_neutral_network #cnn
#deep_learning #facebook #imagenet
فیس بوک با میلیاردها تصویر آپلود شده در اینستاگرام یک شبکه کانولوشنالی را آموزش داد!
فیس بوک اخیرا با استفاده از 3.5 میلیارد تصویر اینستاگرام، در 17000 هشتگ مختلف یک شبکه عصبی کانولوشنالی را آموزش داده است.
این شبکه بر روی داده های تست چالش ImageNet به دقت 85.4 دست یافته است، که بهترین مدل قبل از این به دقت 83.1 دست یافته بود!
نکته قابل تامل نویزی بودن هشتگهای اینستاگرام در مقابل مجموعه داده ی بادقت برچسب زده شده ی آموزشی imagenet است.
به عقیده فیس بوک:
"As training data sets get larger, the need for weakly supervised — and, in the longer term, unsupervised — learning will become increasingly vital," said Facebook. "Understanding how to offset the disadvantages of noisier, less curated labels is critical to building and using larger-scale training sets."
https://www.engadget.com/2018/05/02/facebook-trained-image-recognition-ai-instagram-pics/
https://www.wired.com/story/your-instagram-dogs-and-cats-are-training-facebooks-ai/
#classification #image_recognition
#convolutional_neutral_network #cnn
#deep_learning #facebook #imagenet
Engadget
Facebook trained image recognition AI with billions of Instagram pics
The method could allow deep learning models to be trained with less human supervision.
#سورس_کد
MTCNN C++ implementation with NVIDIA TensorRT Inference accelerator SDK
https://github.com/PKUZHOU/MTCNN_TensorRT
This project can run Pnet and Rnet, the speed is very fast. Just using cmake to build this project
مرتبط با https://t.me/cvision/249
#face #face_detection #face_alignment #MTCNN
#deep_learning #cnn #convolutional_neutral_network
#tensortRT
MTCNN C++ implementation with NVIDIA TensorRT Inference accelerator SDK
https://github.com/PKUZHOU/MTCNN_TensorRT
This project can run Pnet and Rnet, the speed is very fast. Just using cmake to build this project
مرتبط با https://t.me/cvision/249
#face #face_detection #face_alignment #MTCNN
#deep_learning #cnn #convolutional_neutral_network
#tensortRT
GitHub
PKUZHOU/MTCNN_TensorRT
MTCNN C++ implementation with NVIDIA TensorRT Inference accelerator SDK - PKUZHOU/MTCNN_TensorRT
Interactive Tools for Machine Learning, Deep Learning and Math
https://github.com/Machine-Learning-Tokyo/Interactive_Tools#cnn-explainer
https://github.com/Machine-Learning-Tokyo/Interactive_Tools#cnn-explainer
GitHub
GitHub - Machine-Learning-Tokyo/Interactive_Tools: Interactive Tools for Machine Learning, Deep Learning and Math
Interactive Tools for Machine Learning, Deep Learning and Math - Machine-Learning-Tokyo/Interactive_Tools