From Google and Waymo researchers: The self-/unsupervised revolution is near! Unsupervised optical flow model SMURF improves SOTA by 40% and beats many supervised methods such as PWC-Net and FlowNet2

👉 @computer_science_and_programming

SMURF: Self-Teaching Multi-Frame Unsupervised RAFT with Full-Image Warping

Paper:
https://arxiv.org/pdf/2105.07014.pdf

Video:
https://www.youtube.com/watch?v=W7NCbfZp6QE

Code:
https://github.com/google-research/google-research/tree/master/smurf

👉 @computer_science_and_programming

YOLOX: Exceeding YOLO Series in 2021

Anchor-free version of YOLO series

Won the 1st Place on Streaming
Perception Challenge (Workshop on Autonomous Driving
at CVPR 2021)

A simpler design but better performance! It aims to bridge the gap between research and industrial communities.

Paper:
https://arxiv.org/pdf/2107.08430v1.pdf

Github:
https://github.com/Megvii-BaseDetection/YOLOX

👉@computer_science_and_programming

Now removing, duplicating or enhancing objects in video is more realistic with the assist of AI

"We need to talk about the car in the room."
This paper: what car? 🙈

Paper:
https://arxiv.org/pdf/2108.11250v1.pdf

Github:
https://github.com/hustvl/YOLOP

Code:
https://github.com/gist-ailab/uoais#unseen-object-amodal-instance-segmentation-uoais

Paper:
https://arxiv.org/abs/2109.11103

Dataset:
https://paperswithcode.com/dataset/ocid

Project page:
https://sites.google.com/view/uoais

👉@computer_science_and_programming

PASS: Pictures without humAns for Self-Supervised Pretraining

PASS is a large-scale image dataset that does not include any humans, human parts, or other personally identifiable information

Github
https://github.com/yukimasano/PASS

Paper
https://arxiv.org/abs/2109.13228v1

Dataset
https://paperswithcode.com/dataset/pass

Documentation
https://www.robots.ox.ac.uk/~vgg/research/pass/

8-bit optimizers – a replacement for regular optimizers. 🚀, 75% less memory, same with upwards trend, no hyperparam tuning needed Input symbol for numbers: #Lightweight, #LessMemory

Under review as a conference paper at ICLR 2022

8-BIT OPTIMIZERS VIA BLOCK-WISE QUANTIZATION

Paper:
https://arxiv.org/abs/2110.02861

Github:
https://github.com/facebookresearch/bitsandbytes

Video:
https://www.youtube.com/watch?v=IxrlHAJtqKE

Documentation:
https://bitsandbytes.readthedocs.io/en/latest/

👉@computer_science_and_programming

One of the best reference book is definately "Deep Learning with Python" (1st edition) by François Chollet (creator of Keras)

Deep Learning with Python (2nd edition) has been released with 500 pages of code examples, theory, context, practical tips...

Book:
https://www.manning.com/books/deep-learning-with-python-second-edition?a_aid=keras

For online reading:
https://livebook.manning.com/book/deep-learning-with-python-second-edition/chapter-1/

Jupyter notebooks on Github:
https://github.com/fchollet/deep-learning-with-python-notebooks

👉👉@computer_science_and_programming