Open Access Open Access  Restricted Access Subscription Access
Cover Image

An Automated Approach for Detection and Classification Birds and Fish in Still Images

Elham Mohammed Thabit A. ALSAADI, Nidhal K. El Abbadi


Object detection and classification are important task in computer vision (CV) and image processing. when problems of object detection  are meta-heuristic, it treats with distinction the finding of different individual objects in an image. Object classification could be considers as a standard pattern recognition issue. Unlike human who can recognizes any object in the real world with effortlessly,  the computerize recognition of object in image is difficult task.  Detection and classification of animal’s regard as a challenge for the researcher due to many difficulties which reduce the performance and efficiency. This paper suggests to detect the two type of animals (Birds and Fish) in still image, in uncontrolled environment, and classified them as (bird, fish and other (neither bird nor fish)). The proposed algorithm based on using Convolutional Neural Network (CCN). CNN trained by using 3600 different images and tested by using 900 images. Python has used to build the propose model of CNN. The total number of parameters was (1,164,547). We found the best image size for this algorithm is 50x50 and the best number of epochs was 100. The results are promised, the accuracy of testing 900 images reached to 99.99%, with very reasonable time.

Full Text:



  • There are currently no refbacks.