Added sb3-extra-buffers to the project page (#2157)

* Added entry for project sb3-extra-buffers into projects section

* Update changelog.rst

* Fixed title formatting error

Author: Trenza1ore

docs/misc/changelog.rst

@@ -41,6 +41,7 @@ Documentation:
 - Clarify ``evaluate_policy`` documentation
 - Added doc about training exceeding the `total_timesteps` parameter
 - Updated ``LunarLander`` and ``LunarLanderContinuous`` environment versions to v3 (@j0m0k0)
+- Added sb3-extra-buffers to the project page (@Trenza1ore)
 
 
 Release 2.6.0 (2025-03-24)

docs/misc/projects.rst

@@ -277,3 +277,18 @@ various external disturbances and stand up in a few seconds.
 | Github: https://github.com/Rhoban/frasa
 | Paper: https://arxiv.org/abs/2410.08655
 | Video: https://www.youtube.com/watch?v=NL65XW0O0mk
+
+
+sb3-extra-buffers: RAM expansions are overrated, just compress your observations!
+------------------------------------------------------------------------------------
+
+Reduce the memory consumption of memory buffers in Reinforcement Learning while adding minimal overhead.
+
+Tired of reading a cool RL paper and realizing that the author is storing a **MILLION** observations in their replay buffers? Yeah me too. 
+This project has implemented several compressed buffer classes that replace Stable Baselines3's standard buffers like ReplayBuffer and 
+RolloutBuffer. With as simple as 2-5 lines of extra code and **negligible overhead**, memory usage can be reduced by more than **95%**!
+Benchmark results and documentations are on Github, feel free to submit feature requests / ask how to use these buffers through issues.
+
+| Authors: Hugo Huang
+| Github: https://github.com/Trenza1ore/sb3-extra-buffers
+| Relevant project for training RL agents that play Doom with Semantic Segmentation: https://github.com/Trenza1ore/SegDoom