Image size vs. image quality vs. image resolution

Image size vs. image quality vs. image resolution

Image size, quality, and resolution are indicator to describe image properties.

2025-03-17

Machine learning Deep learning

quality resolution size

How to implement multi-GPU training with data parallel?

How to implement multi-GPU training with data parallel?

Distributed main function design, data loader design, move model into distributed environment, calculating metrics by gathering results from different GPUs through redution.

2025-03-17

Machine learning Deep learning

DDP parallel multiple gpus

FLOPs and MACs in deep learning

FLOPs and MACs in deep learning

FLOPs and MAcs are used for measuring the computer performance.

2023-06-26

Machine learning Deep learning

FLOPs MACs

Vision transformer: A new way to analyse image

Vision transformer: A new way to analyse image

Vision transformer slices the image into patches and uses self-attention mechanisms (inter-patches) (qkv interaction) to process the features.

2023-05-24

Machine learning Deep learning

attention mechanism transformer ViT vision transformer

Should I contain background class?

Should I contain background class?

The answer is yes Because most deep learning frameworks for segmentation and detection consider background class (classification task does not have a background label).

2023-05-24

Machine learning Deep learning

segmentation background class detection

PhD thinking accumulation 2023

PhD thinking accumulation 2023

魚沫不見影，水紋又如常。

2023-04-11

Lifetalk

PhD

Experiment experience of deep learning and computer vision

Experiment experience of deep learning and computer vision

This blog has been accumulating my network training and testing experience.

2023-03-29

Scientific Research/科研

scientific research 科研 experience

Good thoughts in 2023

Good thoughts in 2023

There are good thoughts accumulated in 2023.

2023-03-29

Scientific Research/科研

scientific research 科研 thought

Spiking neural network, a bionics way to represent and process data

Spiking neural network, a bionics way to represent and process data

Spiking neural networks are bio-inspired artificial intelligence method to process signal efficiently with low power consumpsion binary signal mechanism.

2023-03-28

Machine learning Deep learning

snn spiking neural network

旅行散记第四期——三月的梅州，是一座充滿花香的城市

旅行散记第四期——三月的梅州，是一座充滿花香的城市

第一次走在梅州的路上，让我感到很特别的是我闻到的花香，一种像洋甘菊般温和的香味。这种香味在各个角落有不同的演绎变化，但都是温和清新的味道。

2023-03-19

Sightseeing

lifetalk travelling soul searching 梅州