Keynes once quipped that public figures who think they are expressing their original thoughts are usually echoing the words of some dead economist. The same might be said of the dead Kant in respect to science. While his thought provides a comprehensive modern framework for science, most practicing scientists have never read him and large swaths of the philosophy of science ignore or reject what they take to be muddled Kantianism.
Bertrand Russell and E.G. Moore were especially hostile to Kant, convicting him of logical errors and supposing that transcendental idealism rested on a mistaken faith in the inviolability of Euclidean Geometry, which Kant presented as the very model of the "synthetic a priori." These dubious charges stuck all the way through logical positivism and continue in the analytic tradition. Certainly Popper's demarcation criteria seem to reject Kantian approaches, or at least so Popper claimed.
Today, however, it seems that more philosophers of science may be open to Kant. Henry Allison offers detailed refutations of the above calumnies, and Kuhn was explicitly influenced by Kant in his anti-Popperian demarcation by "paradigms." He claimed that reading Kant while studying physics utterly altered his naive realism, though those are not his words.
As to the bigger picture. Bacon first described the emerging rift between the rationalists, "the spiders," who weave webs of theory (coherence theory, we might say) and the new empirical naturalists, "the busy ants," who gather bits of data (correspondence theory, roughly). The rift grew into Leibnizian mathematical rationalism and Humean skepticism. It was Kant's great project to merge and mutually limit the two on a firm metaphysical basis, in part to secure the basis of Newtonian physics.
His unique amalgam of coherence and correspondence theory is in many ways, and intentionally, a kind of philosophical version of the hypothetic-deductive framework of science, an expansion of knowledge by rational (conceptual) methods and experimental (intuitional) confirmation. And indeed science does undertake continuous active synthesis based on a priori assumptions of necessity and universality. Even the "experiment" is somewhat Kantian, in that it is hardly passive reception of sense data... the experiential confirmation in "common sense" is quite artificial and actively constructed.
Most working scientists tend to believe they are following analytical rules and discovering "correspondences" with hard "reality." So they look askance at Kantian idealism, regarding it as akin to Berkeleyan empirical idealism or some sort of structuralism. Such assurance was, of course, shaken by statistical mechanics in thermodynamics and then quantum mechanics. But such "shaking" may only register with a few...how many physicists, after all, have time to read Kant? Yet when pressed on theoretical issues many physicists will admit they are constructing models and cannot speak about the "ultimate reality," mere speculative metaphysics, unaware that they are defaulting to a near Kantian position.
Meanwhile, cosmology today seems to be weaving "theoretical webs" well outside any Kantian remit, moving far beyond experimental range and tumbling headlong into the Antinomies. In CPR B511, for example, we see hints of Copenhagen in "you never come face to face with anything unconditioned...." or "...neither a simple appearance [i.e., final particle] nor an infinite composition [i.e., material universe] can ever come before you." One might say Kant attempted to work out a physics that included the observer.
Of course, this is very general and it is easy to contrive this sort of cozy compatibility. I don't know enough about Kant yet to know where he might be in serious conflict with scientific practices, especially in theory of evolution... or perhaps in the evolution and "selection" of theories, where Hegel's criticism of Kant begins. I hope other will offer more specific references to Kant in current philosophy of science, since I'd like to know as well.