Despite of significant achievements made in the detection of target fruits, small fruit detection remains a great challenge, especially for immature small green fruits with a few pixels. The closeness of color between the fruit skin and the background greatly increases the difficulty of locating small target fruits in the natural orchard environment. In this paper, we propose a balanced feature pyramid network (BFP Net) for small apple detection. This network can balance information mapped to small apples from two perspectives: multiple-scale fruits on the different layers of FPN and a characteristic of a new extended feature from the output of ResNet50 conv1. Specifically, we design a weight-like feature fusion architecture on the lateral connection and top-down structure to alleviate the small-scale information imbalance on the different layers of FPN. Moreover, a new extended layer from ResNet50 conv1 is embedded into the lowest layer of standard FPN, and a decoupled-aggregated module is devised on this new extended layer of FPN to complement spatial location information and relieve the problem of locating small apple. In addition, a feature Kullback-Leibler distillation loss is introduced to transfer favorable knowledge from the teacher model to the student model. Experimental results show that APS of our method reaches 47.0%, 42.2%, and 35.6% on the benchmark of the GreenApple, MinneApple, and Pascal VOC, respectively. Overall, our method is not only slightly better than some state-of-the-art methods but also has a good generalization performance.

No abstract

Association rules mining algorithm based on Rough Set theory is put forward using the idea of Rough Set theory, which applies the improved Apriori algorithm in association rules mining on the basis of Decision Table. The advantage of this method lies in three aspects, including the elimination of redundancy attributes, reducing the number of attributes, while scanning Decision Table just once can produce decision attribute sets. Application example analysis shows that this is an effective and fast data mining method.

Reconstruction Method of Network Forensics Scenario has grown into a mature and rich technology that provides advanced skills to get the chain of evidence. Using statistical methods to analyze intrusion logs in order to present evidentiary values in court are often refuted as baseless and inadmissible evidences which is not considering the input spent. These spendings is to generate the reports no matter they are well-grounded evidences or not.Thus, this paper presents the Scenario Reconstruction Method combines the Viterbi algorithm, the most likely sequence of Meta evidence which replaces the Meta evidence was acquired. With suspected evidence, thus obtaining the chain of evidence. However, the Viterbi algorithm parameters is derived from the Baum-Welch (B-W) algorithm, and the B-W algorithm is easy to fall into local optima solution. While an Adaptive Genetic Algorithm (AGA) is used to estimate parameters of the Hidden Markov model (HMM), where Chromosome coding method and genetic operation mode are designed. The experimental results show that, this method can accurately reproduce the crime scene of network intrusion, compared with the network forensic evidence fusion method which is based on the HMM. The method has been applied to forensics system, and has obtained good result.

No abstract