Segment Anything 2 (SAM2) from Meta: The next generation of Meta Segment Anything Model for videos
Segment Anything Model or SAM was introduced a year ago by Meta. The model was revolutionary and proved that a single model can be used to segment any objects in images.
Meta has now come up with SAM2 which extends SAM to videos. In this video I deep dive into the SAM2 model, the data engine to generate the largest video dataset to date - the SA-V dataset. I also do a quick walk through of the experiments and results.
⌚️ ⌚️ ⌚️ TIMESTAMPS ⌚️ ⌚️ ⌚️
0:00 - Intro
1:13 - Challenges with video segmentation
1:41 - Overview of SAM2
2:58 - Promptable Visual Segmentation
4:42 - SAM2 Model
5:19 - End to end architecture
5:32 - Image Encoder
5:45 - Memory Encoder
6:19 - Memory Bank
7:01 - Memory Attention
7:20 - Training
8:03 - Data Engine
10:54 - Segment Anything Video (SA-V) dataset
11:23 - Experiments
AI BITES LINKS
YouTube: / @aibites
Twitter: / ai_bites
Patreon: / ai_bites
Github: https://github.com/ai-bites
WHO AM I?
I am a Machine Learning researcher/practitioner who has seen the grind of academia and start-ups. I started my career as a software engineer 15 years ago. Because of my love for Mathematics (coupled with a glimmer of luck), I graduated with a Master's in Computer Vision and Robotics in 2016 when the AI revolution just started. Life has changed for the better ever since.
#machinelearning #deeplearning #aibites