There are various pre-trained deep learing models availabel which are trained using data present in ImageNet(or any other data source) like AlexNet, VGGNet, Inception, ResNet, Xception, MobileNet, SqueezeNet etc. I want to develop a certain object detection application which will be ultimately sold to the customer. I have few questions as follows:
Can we use these pre-trained deep learing models for commercial use direclty?
If object of interest is not supported by these models, can we use their architecture and perform learning using our own data set and use it for commercial use.
All Source code in GitHub has a license. The License explains whether the code can be used or not. If used then under what circumstances and whether and what you owe to the creators when implementing the code for your science project or as a product that you want to launch.
Some are free and has no problems with it being modified and used where others are more protected.
Please go through the different licenses usually used and understand what are the key points of the license on the code you want to use to avoid legal disputes in the future.
In the case where the license does not care what you do with the code..both the answers to your above 2 questions is yes..