Suggesting a technical solution is a reaction to an outcome; the solution might not address the contributing factors thus further problems may arise.
A takeoff performance monitor based on acceleration could (with additional risk) detect a low thrust setting, but it is unlikely to detect an incorrect speed setting for weight and/or configuration.
It would be sensible to review what has changed; has the frequency (rate) of this type of incident actually increased? If so then what changes have there been which might contribute to the issue.
The obvious candidate is technology, but this also involves human interaction. Technology should be able to provide a very high level of error detection / prevention; apparently a sufficient level was achieved with previous methods, e.g weight and speed cards. Did previous methods have intrinsic defenses or generate specific crew skills, e.g. never showing ZFW on the card (irrelevant), or routine card use provided a mental pattern of the normal range of speed / weight / thrust combinations – a rule of thumb check.
Similarly, suggestions for ‘airmanship’ qualities are of little value unless the specific qualities are defined and the mechanism of defense explained. From these it might then be possible to train and improve these features. Unfortunately, even with excellent human qualities everyone can suffer inadequate performance due internal or external influences, and currently we can neither identify nor control these.
Perhaps a greater concern is that outside influences are affecting the human interaction with technology, e.g. shorter turnaround times increase the pressure to rush tasks.
Have turnaround times have been reduced because tasks have been automated, but then these tasks suffer from the tendency to rush, etc, etc. What other factors may contribute to current problems, what else has changed – human, technology, environment, society?
The industry strives to improve safety; this will probably have to involve both the human and technology. However, the use of technology should not trigger additional pressures on human behavior; any freed-up time could be used to ‘create safety’. Alternatively we could consider using a mixture of the old and new; instead of crosschecking the output of two computers (garbage in/garbage out), use speed cards as a cross check – and you might need these if the computers are on the MEL.