Using this repo for a Tiger recognition task?

[nhprasan]

Namaste,

I’m considering using this repository as a starting point to build an individual tiger recognition system for a tiger census. My plan is to implement a Siamese network architecture, but I’m uncertain about how to handle modifications. I'm thinking i could use the megadetector in this repo to detect tiger and then run it through a siamese network architecture built on CNN (i.e., the classifier CNN in this repo) which is trained on wildlife dataset to extract features.

Additionally, I noticed that the dataset used for training the mdoels is restricted to "Amazon rainforest" which doesn't have any tiger population. So I presume that the model will need to be fine-tuned with a tiger dataset first. Am i correct or are the models also trained on a wide dataset that includes detection of tigers?

I believe the idea is feasible, but I’m seeking guidance on how to modify the approach to address challenges and successfully implement the system. Could you provide some insights or suggestions on the necessary adjustments? I may be having some blind spots in approaching this problem and any guidance on this would be helpful.

The work we intend to do will be open source and for the benefit of anyone who would like to make use of it.

Thank you.