licensing issues

[ispysoftware]

💡 Your Question

I'm looking to get this integrated with CodeProject's AI server but in the license you say:

"You shall not, without Deci's prior written consent: (i) resell, lease, sublicense or distribute the Software to any person; (ii) use the Software to provide third parties with managed services or provide remote access to the Software to any person or compete with Deci in any way; (iii) represent that you possess any proprietary interest in the Software; (iv) directly or indirectly, take any action to contest Deci's intellectual property rights or infringe them in any way; (V) reverse-engineer, decompile, disassemble, alter, enhance, improve, add to, delete from, or otherwise modify, or derive (or attempt to derive) the technology or source code underlying any part of the Software;"

So this software is undistributable and unmodifiable?

Versions

No response

[dagshub[bot]]

Join the discussion on DagsHub!

[Louis-Dupont]

Hi @ispysoftware YOLO-NAS is available under an open-source license (Apache -2.0) with pre-trained weights available for non-commercial use on SuperGradients as you mentioned. We can allow distribution and modification as long as there is no commercial use in weights or derivatives, including in connection with any models used in a production environment. Note that the rest of the SuperGradients, and the other pre-trained models (YoloX, PPYoloE, ...) are all free of commercial use (Apache -2.0)

[trathpai]

Thankyou for great work. @Louis-Dupont If pretrained weights are used for transfer learning on custom dataset , does the usage of resulting weights in commercial work violates the copyright?

[Gabriellgpc]

@Louis-Dupont , just to clarify / double check, YOLO-NAS can be used with commercial purpose (Apache - 2.0) if , and only if, we train it from scratch (avoiding using pre-trained weights)?

[Ivansstyle]

Hi, cannot see above questions answered anywhere else. Same question here: in what exact terms model is suitable for commercial use: when trained from pre-trained weights? When trained from scratch? Thank you again for amazing work!

[giorgiounity]

Hi @Louis-Dupont , I need the same clarification asked by @Gabriellgpc . Can you clarify if YOLO-NAS can be used for commercial purposes (Apache - 2.0) if it is trained from scratch?

[lherman-cs]

+1 I need the same clarification wether if YOLO-NAS can be used for commercial purposes if it is trained from scratch

[harpreetsahota204]

Hi @lherman-cs @giorgiounity @trathpai @Ivansstyle @Gabriellgpc @trathpai

The commercial use restriction applies to the pre-trained weights and any downstream model that is create using the pre-trained weights.

You can use the architecture freely if you are training YOLO-NAS from scratch.

[BloodAxe]

Possible duplicate of https://github.com/Deci-AI/super-gradients/issues/894 https://github.com/Deci-AI/super-gradients/issues/901 https://github.com/Deci-AI/super-gradients/issues/983

[BloodAxe]

As @harpreetsahota204 clarified - The commercial use restriction applies to the pre-trained weights and any downstream model that is create using the pre-trained weights.

[NeovMs]

If we use the pre-trained models for knowledge distillation to pre-train the Yolo-NAS from scratch, we're not strictly using the weights. In this case, can we get round the licence ?