Ask IH: Your SaaS had an insurmountable technical problem - what did you do?

Hello indies,

I've chiselled away at my SaaS since late 2013. It "turns PDFs into data", and the main USP is detecting tables. V1 (and the entire competitive landscape) works well for tables that have lines, so I'm working hard on V2 (deep learning).

I met a PhD Computer Vision expert who is interested in helping and get a stake in the company, but things have been slow-going (1 year since they came on board, with catch-ups once a month). I've since completed the deeplearning.ai specialisation and gave the problem a stab, but you can imagine how little progress I'm making considering that I have 2 small kids, and have to contract to keep the family afloat.

I'm still trying to find clients for V1 - even if the core doesn't have a 100% success rate, there is a market for my offering. BUT, there isn't much joy in peddling a clone or a copycat product.

What to do?

I thought about opening up the challenge to the greater community - "solve this problem and win a stake in the business".

The other option is to put the job on Upwork or somesuch, but I have a feeling the really good scientists are working on more exciting stuff that PDFs.

I'd be interested to know if other indies on here had an insurmountable technical problem, and what they did to solve it.

Thanks,
Juan