College English classroom teaching evaluation is one of the key issues discussed by all schools at present. First of all, teachers and students are highly concerned about English learning in the compulsory course of English examination, but there are also many problems. The distance network teaching with computer technology as the main body develops rapidly, a large number of video and audio teaching resources network through the network transmission to present in front of learners, through the network transmission of video and audio aided, expand the audience of network teaching, is conducive to the realization of digitalization, information, lifelong, new education goals. Firstly, this article summarizes the problems of poor effect, uneven performance, and mismatch between evaluation and teaching in college English at present. Then, based on the previous works, the author designed a machine learning-based Internet of things technology and verifies its feasibility. The data of teaching experiments show that the writing performance of the students in the lower section of the experimental class is better than that of the students in the middle and high sections. Among them, significant progresses have been made in grammar, length, expression, and structure, which has optimized students’ preclass preview efficiency, autonomous learning motivation, quality of homework completion, and afterclass reflection behavior. Finally, the author summarized the shortcomings of this research and puts forward the prospect of relevant research, in order to provide reference value for the national college English blended teaching and promotes the efficient implementation and vigorous development of college English teaching. In the future, the evaluation index system under the pressure of performance can be further studied.