A complete list or fraction will be subjective, because each case has nuances - for one thing even languages with definite articles use them differently - and because it is always subjective what should count as a separate language.
Roughly speaking, among the languages of Eurasia and the Western world, essentially Slavic, Ural-Altaic (regardless of the dispute around it), Indo-Aryan, Dravidian and Berber languages can have definite NPs by default or no articles at all, including for countable nouns. The oldest forms of Greek and Latin had no definite article, but they later developed them. Please pardon my ignorance on the rest of the world.
It seems to be a partly areal feature, for example it includes the Eastern Indo-European languages that had long and intimate contact with Altaic and Dravidian languages.
That said, no matter the counting methodology used, default definite NPs ie implied definite articles are rare, much rarer than languages with no articles at all. Furthermore, the overall distribution seems to be consistent between Eurasia, Austronesia, Africa and the New World.
To understand the nuance, let us examine some of well-known languages with NPs that are definite by default. In formal Persian it is so:
In the literary language, no definite article is used; rather, it is implied by the absence of the indefinite article.
In Turkic languages it is similar, with some nuances. For example:
Araba sokakta. (The car is in the street.)
Breaking it down a bit, we have Car street-IN. Araba simply means car, and is in its pure nominative and canonical form. (You can regard -ta and its variations either as a locative case ending or as a postposition meaning in, for our discussion it does not matter.)
Furthermore, notice that sokak (street) is also a definite NP.
So this is a good example of default definite interpretation of NPs.
In both language groups, however, there are many cases of countable, and of course uncountable, NPs without any article that would be translated into, say, English with the indefinite article or with no article, for example:
آیا اتومبیل دارید؟ | ("Do you have a car?")
Kar var mı? | ("Is there snow?")
Only in some context would one interpret this as Do you have the car?, and it is hard to dream up a context where "Is there the snow?" would make sense. Thus in such instances it functions more like a language without any articles.
Compare this to the following, with the indefinite article is strange or emphatic, a bit like English one (as opposed to a/an):
آیا یک ماشین دارید؟ | ("Do you have one car?")
Bir kar varmı? | ("Is there one snow?")
Again, it only makes sense in context or almost never.
In Slavic languages, there are generally neither definite nor indefinite articles, which is actually very common among the world's languages. In these cases, the default interpretation is mostly a question of the context, although naturally native speakers do not bother to map every NP to the concepts of definite and indefinite in an English speakers' mind.
As you are interested in the overview, some exceptions to the groupings: among Iranian languages, Ossetian and informal Persian and Kurdish can mark definiteness explicitly, and apparently Tajik marks only objects, like Turkic languages. Hungarian has very simple and explicit definite and indefinite articles, like Western European languages. Among Slavic languages, Macedonian and Bulgarian and related dialects do have a definite article, like other Balkan languages. Again, this is definitely an areal feature. Also, I cannot find a colonial-era creole that does not have a definite article.
In any case, there are always ways to be explicitly definite, eg with demonstratives, ie by saying that frog, and indeed definite articles in many languages were derived from such ways. To me anyway it is not clear where to draw the line.
